up-to-date electronics for lab and leisure \ / % / \ \ i % * tw '* V ■ A Jm* * tv sound modulation systems how to gyrate , v > * l ♦ 41 Febnjary • • co ntents elektor february 1975 — 203 14 mu § % W minidrum Elektor laboratories have designed a new electronic instrument which shows great similarity to a real set of drums. In future issues "sw instruments end an automatic rhythm section will be described. philips sonant in the snide 'sonant”, elsewhere in this issue, a definition is given of a new type of "active” sound- reproducer = = : - practical realisation of the sonant-principle is to be found in the recently presented Pi-' ps 22 - - 232 zlectronic". This article describes a few aspects of the way in which this sonant operates. motional Taedback ■ piai improving readability of displays . , 1 - i i i microdrum L - “ jc more or less with the minidrum published elsewhere in this issue of Elektor, this article describes the dopiest member of the electronic drum family. d . ! i t s r s a i u i s p . ~ i y ■ ■ ■ ■ . ■ . . . ■ ■ * ■ ■ ■ > ■ ■ . . ■ ■ ■■ p . * . # ^ . D . ■ . B , a a a D m M CJ a a a a a z t tor laboratories have designed a universal display which should satisfy the requirements of most enthusiasts. Tpb display may be used with seven-segment indicators of the LED or Minitron type. # S * ■ i i A i ft # dil ied probe Tne probe discussed in this article is suitable for testing 14- and 16-pin dual-in-line ICs under operating condi- tions. the missing Hink I P I i elektor "sonant" loudspeaker system An earlier article explained the operating principle of the electronic loudspeaker. It was shown that this ap- proach enables a woofer in a relatively small enclosure to provide really good reproduction quality from 40 Hz upwards. The second step is to describe a complete system, using active crossover networks, based upon the electronic loudspeaker used as a woofer. I -S 4 m a ft ■ tv sound , , , , . , . For those who are dissatisfied with the quality of their television sound (and who isn't?) here is a design which enables a high-quality sound signal to be derived from a television receiver and fed into a separate amplifier and speaker system without making any connection to the T.V. set. three-eyed bandit . From time immemorial man has played games of chance. Modern technology has considerably widened the scope of games of chance so that a whole range of 'gaming machines' can be seen today. 4ift4ftlh4»4»4i * big ben 96 Nowadays both mechanical and electronic doorbells playing complete melodies are commercially available. The Big Ben plays a striking end well-known melody, cos/mos digital ic's mos tap Only recently has rt become possible economically to replace the simple electromechanical switch by an electronic system with no moving parts — the Touch Activated Programmer (TAP). 4 # » 1 i * 4 1*11 modulation systems This article should give an insight into the various modulation systems nowin use and will also explain the de- sign philosophy of a new transceiver developed by Elektor. maxi display recip-riaa The performance of preamplifiers for magnetic pickup cartridges is invariably not sufficiently well known. The weighting network described in this article greatly simplifies measurement of the amplitude response (RIAA or I EC curve}, overdrive margin, distortion, signahto-noise ratio and hum level. how to gyrate - and why “he gyration principle was suggested by theoreticians over 25 years ago, although it is rarely seen in practical circuits. In this article the theoretical principles of the gyrator are discussed, and some practice! circuits and applications are presented. improved 7 segment display psychedelic lights 208 223 227 234 238 239 241 246 255 208 — elektor february 1975 minidrum minidrum The Elektor laboratories set out to design a new electronic instru- ment which would show great similarity to a real set of drums. The success of this project was con- firmed by a number of percussion players at a Hi-Fi exhibition in Amsterdam, where demonstrations of the Minidrum proved highly successful. The great advantage offered by a con- ventional set of drums over an automatic rhythm generator is that the player produces all the rhythms himself, thus resulting in a more lively and varied sound. Of course, the novice percussion player will initially experience difficul- ties in producing different rhythms, but these will come with practice. The three percussion instruments gener- ally used in a basic drum kit are bass- drum, snaredrum and cymbal and the Minidrum consists of these three instru- ments. In future issues the system will be expanded to include several more instru- ments, a ruffle system and an automatic bassdrum for keeping time. Finally we shall publish a design for a fully auto- matic rhythm generator for those who want this facility. Design of the Minidrum The Minidrum is controlled by a touch activated programmer (TAP) instead of by mechanical pushbuttons, and com- prises four boards (see figure 14) the TAP board, gyrator boards for the bass- drum and snaredrum, a noise board for the cymbal and snaredrum and finally a mixer-preamplifier. The TAP The TAP circuit of figure 1 utilises an RCA COSMOS hex-inverter IC. This is available in two versions, the CD4009AE and theCD4049AE, the difference being that the 4009 has an extra supply con- nection to the output stage (pin 16). Production of the 4009 is shortly to cease as it has been found that this con- nection gives trouble in certain applica- tions. Figure 1 shows the connection for the 4009, where a protection diode D] is connected between positive supply and pin 16. If the 4049 is used D, may be omitted as pin 1 only is the positive supply connection. The operation of the TAP is very simple and since the three channels are identical the Bd channel only will be described (Bd, Sd and Cy stand for bassdrum, snaredrum and cymbal respectively). In the quiescent state the input of I, is held high by R, . The output is thus low and C] is uncharged. If the Bd input contact is touched hum from the skin causes the output of i! to switch between ‘0’ and ‘1’ at 50 Hz (I j is actually operating as a very high input impedance, high gain amplifier). C! charges via D 2 and the voltage at the Bds output is used to control the bassdrum. When the contact is released Cj discharges via R7 into the bassdrum circuitry in about 60mSec. R7 is included to limit the current into the bassdrum circuit and thus to protect both the bassdrum circuit and the out- 1 put of Ij. The TAP p.c. board The TAP p.c. board is shown in figure 2 and the associated component layout in ] figure 3. It can be seen that there is space for six TAP channels on this board, but j only three of them are used in the basic Minidrum. The others will be used in the more sophisticated systems in future issues. When constructing this board it is re- commended that a socket is used for the 1 Components list to figures 1 and 3: resistors: R 1> R 2- R 3 = 27 M or 10 M r 7- r 8' R 9 = 47 k capacitors: Cl.C2.C3 = 0,22 jU C7 =100/i, 10 V semi-conductors: IC= CD4009AE or CD4049AE (RCA) Di,D 2 ,D 3 ,D4= DUS Parts are coded in conformity with codes on the universal PC board designed for a maximum of six inputs and outputs. +Vb=6.8 V/1 mA produces the bassdrum and snaredrum sounds is given in figure 4. As the gyrator may be an unfamiliar concept to some readers the theory is discussed elsewhere in this issue, but a few words of explanation will be given here. In essence, by using a gyrator a capacitor may be made to simulate an inductor. Bulky conventional inductors may thus be eliminated in certain applications. In figure 4 C 6 is gyrated into an inductor and this simulated inductor appears across C^ thus forming a parallel-resonant circuit which determines the frequency of the instrument. The gyrator board has two control in- puts, input 1 and input 2, each com- prising a monostable and filter. Only one of these is shown in figure 4, however, as only one control input is used in the basic Minidrum. The monostable consist- ing of T! and T 2 receives the control signal from the TAP and produces a rectangular pulse. This is shaped by the filter consisting of C 3 , C 4 , C s , R I0 and R n to give the appropriate attack and decay characteristics for either the bass- drum or the snaredrum, depending on the component values. This shaped pulse is applied to the base of T 7 to control the gyrator. Output P' is not used in the bass- drum, but in the snaredrum it is connect- ed to the noise board. The output of the gyrator if filtered by C 8 , C 9 and R 23 . The gyrator board The printed circuit for the gyrator is given in figure 5 and the component lay- outs for the two versions are shown in figures 6a and 6b. Note the differences between the two boards where compo- nents are omitted in the snaredrum version and also the space for an addi- tional monostable at the top left-hand corner of both boards. The noise circuits The conventional snaredrum, in addition to the basic tone, produces a character- istic ‘tizz’ caused by the catgut snares Figure 1. The circuit of the basic Minidrum TAP. Figure 2. The TAP p.c. board which has pro- vision for up to six TAP channels. Figure 3. The component layout for the TAP as used in the basic Minidrum. As only three inputs are used the component codes do not all follow in numerical order since the board is coded for six channels. IC to avoid the possibility of damage due to static charges, leakage from unearthed soldering irons etc. The IC should be the last component mounted on the board. If the input leads from the touch con- tacts are longer than about 3 cm. then they should be screened with the screen- ing connected (at one end only) to the supply common. The output leads need not be screened if they are shorter than about 6 cm. Bassdrum and snaredrum gyrator The circuit diagram of the gyrator which GYRATOR GYRATOR minidrum 210 — elektor february 1975 +Vb2=6.8V/2mA 6a | — 1 + o *Vb2 (+>—«• Components list to figures 4 and 6. resistors: Rl,R 2 r 3- r 4 R 10 R 11 = 10 k (10 k) = 470 k (470 k) = 4M7 (4M7) = 4k7 (100 k) 6^ O + Vb2 0— • Ur+l t T T f ’ «:JL **±»*i. 0— Rl 2 toR 21 = 6 k 8 ( 6 k 8 ) R 22 - 27 k (470 k) R 23 = 470 k capacitors: Ci = 150 n (18 n) C 3 = 10 n (omitted) C 4 = 33 n (10 n) C 5 - 27 n (10 n) Cg - 1 /inon-electrolytic (56 n) C 7 - 330 n (150 n) Cg - 1 /Ltnon-electrolytic (100 n) Cg - 1 00 n (omitted) C 10 = 100/X 10 V semi-conductors! Dl.D 2 .D 5 - DUS t 1 ( t 2 =tun t 5- t 7- t 9 = BC107B, BC108B, BC109B T 6 ,T 8 = BC1 77B, BC1 78B, BC1 79B Component values in brackets apply to the snaredrum gyrator, those without brackets to the bassdrum. minidrum elektor february 1975 — 211 Figure 4. The circuit of the gyrator as used in the basic Minidrum for the bassdrum and snaredrum. The component values in brackets apply to the snaredrum, the others to the bassdrum. Figure 5. The universal gyrator p.c. board. Figure 6 a. Component layout of the bassdrum gyrator. As can be seen the components relating to input 2 have been omitted. Figure 6 b. The component layout for the snaredrum gyrator board. Again the compo- nents for input 2 have been omitted. Figure 7. The noise circuitry for the snaredrum and cymbal. As can be seen from the diagram the cymbal is driven from a monostable on the board. +Vb2=68V/2mA Components list to figures 7 and 8 . resistors: R 56- R 68' R 71- R 76 = 100 k r 57 - 2M7 r 58- r 60- r 61- r 69- r 70- r 75- r 97 = 10 k R 59 = 4k7 r 62- r 63- r 65- r 73- R 9 2 = 470 k R64 = 820 k r 66- r 93 = 6k8 Rg 7 = 330 k R 67a R 94a = 10 M R 72- R 98 = 27 k r 74> r 77 “ 270 k R 94 = 680 k R 95 = 120 k r 96 = 8k ® Po - 10 k, preset capacitors: C 16 = 100 M, 10V C22 = 47 n c 23- c 28 = 4n7 C24 - 1 50 n C 2 5 = 120 n C26 = 1 2 n C 2 7 = 220 n C30 = 100 p c 31- c 32 = 1° n C-iq = 8n2 C40 = 22 n C 41 - c 42 = 2n7 ifo.[ \ *>^^026 _.'r Cymbals +Vbi=12V ....17V/ 2mA 44 . semi-conductors: t 17. t 18. t 21- t 22 T23- t 26 =TUN t 19- t 20- t 27. t 28 =tup D 7 , Dh.Dt 2 .Di 3, Di4,Dig,D20.D21 = DUS ® * minidrum 212 — elektor february 1975 stretched across the drumhead. This is simulated in the Minidrum by filtered noise from the noise board, the circuit of which is given in figure 7 . The cymbal, on the other hand, has no basic tone, but is simply filtered noise, which is also de- rived from the noise board. The noise circuits operate in the follow- ing way. Noise is generated by T 2 i (about 5mV appears at the emitter) and is ampli- fied by T 22 and T 23 . The noise amplitude may be adjusted between 0 and 3 V by means of P 2 . Noise for the snaredrum is applied to the junction of R94 and R95 via C 41 . T 27 is normally cut off which means that T M is also cut off and the noise signal is blocked. When the snare- drum gyrator is activated a pulse from the P’ output appears on the input (C39). This is differentiated and the negative going trailing edge briefly switches on T 27 , thus rapidly charging C40 • T 28 is thus biased on and the noise is amplified by T 28 and T 26 until C40 discharges and the voltage on the base of T 28 falls below about 2 V when the transistor cuts off. The effect obtained is thus a ‘thump’ from the gyra- tor followed by the ‘tizz’ from the noise board. The cymbal noise circuit operates in exactly the same manner but it has its own monostable on the noise board driven directly from the TAP since the cymbal has no associated gyrator board. T 19 and T 2 o perform the same function as do T 2 7 and T 28 in the snaredrum noise minidrum elektor february 1975 — 213 scflMII Figure 8a. The circuit board for the noise circuitry. Figure 8b. The component layout of the noise board. The components for the monostable have been omitted in the snaredrum input. Figure 9. A photograph of a completed noise board. Figure 10. The simple power supply used with the Minidrum. The transformer should be able to supply about 100mA. at 9-12 V RMS. Figure 11. The prototype of the Minidrum constructed at the Elektor Laboratories. Figure 12. A mixer-preamplifier which may be used with insensitive power amplifiers. The gain is adjusted by means of Py. Figure 13. The board and component layout for the mixer-preamplifier. Components list to figures 12 and 13. resistors: R 2 4 = 22 k R 25- R 40 = 10 k r 26 = 27 k r 34- r 41 =470fi r 35- r 36 = 150 k R37 = 680 ft 8k2 220 k, preset bassdrum snaredrum 2 semi-conductors: Tio-Tii.t 12 = tun D 6 = DUS 214 — «l»k tor february 1975 minidrum Figure 14. Interconnection diagram for the complete Minidrum. Figure 15. The complete Minidrum mounted in a perspex case. The copper furniture tacks used for the touch contacts can be clearly seen. circuit. The cymbal noise is mixed with the snaredrum noise at the junction of Rgs and C 22 • The noise board The noise board of figure 8a and b has provision for a monostable on both in- puts. These will be used in ‘noise’ instru- ments described in future issues. The snaredrum however requires no mono- stable as it is driven from the gyrator board. This monostable may therefore be omitted and links soldered in as shown in the component layout of figure 8b. Mixer preamplifier If the power amplifier to be used with the Minidrum has sufficient sensitivity for the 50mV output produced by the instru- ments then the outputs of the three boards may simply be linked together and fed into the amplifier. If this is not the case the mixer-preamplifier of fig- ure 1 2 may be used. In this case resistors R 22 on the gyrator boards should be re- placed by a wire link. The mixer pre- amplifier is a simple two stage voltage- amplifier with virtual earth input and a low impedance emitter-follower output. P! is included to adjust the gain. The p.c. board and layout are given in figure 1 3. It will be noted that there are 1 0 inputs on the board but only 3 of these are used in the basic Minidrum. Power supply The Minidrum is fairly insensitive to interference so a sophisticated power supply is not required under normal domestic conditions (e.g. not in close proximity to heavy electrical machinery) and the circuit of figure 10 will be quite adequate. The +6.8 V supply for the in- struments and the TAP is derived from a simple zener stabilizer D 5 and the un- regulated supply for the mixer-preampli- fier is taken directly from Ci . The circuit may easily be constructed on a piece of stripboard. Construction +Vb2 +Vb1 is a matter of personal preference and the photographs of figures 1 1 and 1 5 are merely a guide as to the type of layout. The prototype was mounted in a perspex box for visual purposes but from an electrical point of view a metal case is desirable for screening purposes. In the photograph the TAP board is on the right with the noise board directly behind it. The two gyrator boards are mounted one on top of the other at the bottom left and the power supply is at the top left. The main points to remem- ber in the construction are that the connections to the touch contacts should be as short as possible, as should signal connections between boards. Screened leads may be used if necessary with the braiding connected (at one end only) to the supply common. Copper-plated fiimitnrf* tar*lrc arp iHpul fnr thp toiirh contacts as they are easy to solder and practically immune to oxidation. The panel carrying the contacts must be of an insulating material such as perspex or paxolin. For ease of construction the Minidrum may be assembled using boards available from Elektor. M sonant elektor february 1975 — 215 sonant One of the weaker links in the sound reproduction chain has always been the loudspeaker. The physical problems involved in loudspeaker design are more severe than those occurring elsewhere in the chain — but this is no reason for failing to face them. A loudspeaker can be defined as a ‘more or less linear convertor of electrical to acoustical power’. One can now attempt to make detail improvements to the con- version mechanism (flattening its power- frequency response or reducing its non- linear distortion etc.), or one can look for a different approach to the entire conversion problem (e.g. electrostatic drive!). Still another approach is perhaps more fundamental: combine a power amplifier, a transducer (loudspeaker) and an error-correcting system to form a ‘linear convertor of electrical voltage to acoustical amplitude’. This approach is sufficiently new to merit a new name. We suggest the name ‘Sonant’. An example of such a design is the 22RH532 Electronic, recently introduced by Philips (see the article ‘Philips Sonant’ elsewhere in this issue). The introduction of a low-distortion voltage-to-sound convertor would mean that all the links of the sound repro- duction chain reached an acceptable level of performance. It seems desirable to do a little rethinking about the whole sound reproduction process, in the light of the changed situation. Figure 1 shows a block diagram of a re- production chain using sonants. The fig- ure also shows the influence of various links on the frequency characteristic, stereo balance, stereo width and ‘0 dB’ reference level. The influence on the level is important when a loudness-contour correction is to be used. The reproduction chain includes a ‘pre- sonant’. This is the link which applies the corrections necessary to attain the desired overall characteristics in the face of errors elsewhere in the chain. Essen- tially this function can be performed by a normal control-amplifier. Optimum convenience of operation, however, re- quires some rather different control arrangements: — tonal balance. Assuming that the recording staff have done their job properly, it should only be necessary to make tonal balance adjustments which will offset the effects of the listening-room acoustics on the fre- quency characteristic. A once-only adjustment of preset controls is then sufficient. — stereo balance. This is somewhat de- pendent on the positions of the so- nants in the room. A preset adjust- ment will once again suffice. — stereo width. This is also somewhat dependent on the sonant positions. It is furthermore very dependent on the personal taste of the recording engineer, so that a control provid- ing some range of variation is desir- able. Experience has taught that four switched conditions are ad- equate: mono — ‘half stereo’ — stereo — ‘wide stereo’. — level. The requirements to be met by the contour compensation depend on the actual reproduction level at any given moment. However, the ‘0 dB level’ may be 1 00m V for example from a disc preamplifier, 300mV from an FM tuner and lOOOmV from a tape recorder. It is therefore necessary to match levels from the various inputs. The final requirement is for a volume control fitted with the loudness-contour com- pensation circuit. Once again experi- ence teaches that the number of set- tings can be drastically pruned. What about: background — moderate — normal music reproduction — shatter- ing? It will be clear that most of the front panel controls disappear. These were in the past used mainly to compensate for imperfections in the recording- or play- back-equipment. The operating pro- cedure now becomes much simpler: ‘se- lect programme’, ‘select loudness’ and (for the time being?) ‘select stereo width’. The well-known complexities of operating audio equipment have now been replaced by the ease of operation of a TV set (once properly adjusted!). Design work is at the moment in progress on a complete elektor-presonant, along the above lines. u 216 — elektor february 1975 philips sonam In the article "sonant ", else I 1 1 I I L/ | I Q | | I where in this issue, a definition ■ ■ is given of a new type of "active" sound-reproducer design, in which the radiated sound (pressure) level is linearly dependent upon the electrical signal (voltage) applied to the input. This approach differs fundamentally from the classic 'passive' loudspeaker system design, where electrical input power is con- verted - with limited linearity - into radiated acoustical power. Passive systems are either particularly poor in low-frequency linearity or else they are very bulky. On the other hand, active systems with good low- frequency performance can be quite compact. A practical realisation of the sonant-principle is to be found in the recent- ly presented Philips 22RH532 "Electronic". This article describes a few aspects of the way in which this sonant operates. It is well known that the performance of most loudspeakers is far from ideal. The designer of a complete system is invariably forced to make compromises between size, shape, efficiency, direction- ality, distortion, amplitude response ... and price. The attempt to achieve really good performance via the classic ap- proach nearly always fails. The perform- ance in the bass register, in particular, only becomes acceptable as the enclosure or system becomes too large for dom- estic listening-room convenience. The possibility now suggests itself of circum- venting the mechanical-acoustical prob- lems by electronic means. One possible approach is to make use of a power am- plifier with a negative output impedance, which can compensate for some of the drive-unit’s DC resistance and so obtain better control of the coil movement. Examples of this approach are the “Electronic loudspeaker” (Elektor so- nant) and a commercially-available sys- tem called “Servosound”. The difficulty with this approach is that the systems have to use current-depen- dent positive feedback around the power amplifier, which means that the optimum adjustment is fairly critical. A different approach becomes possible when the loudspeaker is equipped with some form of motion-sensing device. One can then apply overall negative feedback to the complete amplifier-driver system: so- called ‘motional feedback’. This approach is what has been realised in the wooler channel of the Philips sonant. It should be noted that neither of these ideas is new. The negative-output- impedance approach was described as far back as 1940, by nobody less than Harry E. Olson. The motional feedback approach using an accelerometer was first tried at least ten years ago. The difficulty was to design a reliable system, not too expensive - and suitable for mass production! Principle Figure 1 is a block diagram of the com- plete sonant. It is clear that motional feedback is onlv aDnlied to the woofer channel, the classic approach being adopted for the mid-range and the treble. In contrast to the “electronic loud- speaker”, this design employs a single amplifier for the mid- and treble-ranges, in conjunction with a passive dividing network. The bass register (35 -500 Hz) is handled by the feedback system. This subchannel consists of a 40 watt amplifier, the bass transducer with accelerometer and a feedback network. The feedback net- work actually contains an impedance- matching circuit for the piezo-electric acceleration-pickup, a preset gain control and a set of stabilising filters. Before examining the system in detail it will be interesting to see what results are actually achieved in practice. Results The volume of the actual woofer en- closure is only 1 5 litres, the overall out- side dimensions being 28.5 x 38 X 22cm. A copy of Elektor, folded out flat, is therefore rather larger than the system’s front panel! The complete electronic ‘works’ (amplifiers, filters, power supply) are mounted inside the ‘box’, as can be seen in the photograph (figure 2). In a small enclosure such as this the fundamental resonant frequency is about 80 Hz, but the feedback arrangement prevents this having any effect on the amplitude response. This response- characteristic is sketched in figure 3. The 3dB rolloff-points are shown at about 35 Hz and 20 KHz. The dotted curve shows what happens when the feedback is made inoperative — more than an octave of bass response is lost! A further advantage of the use of feed- back is that the distortion is reduced. This is shown in figure 4. One should bear in mind, at this point, that the subjective effect of loudspeaker distortion is quite different to that of the usual kind of distortion in the (power) amplifier. A good amplifier is substantially free of perceptible distortion until it is actually "Motional Feedback" System philips sonant elektor february 1975 — 217 Figure 1. Block diagram of the Philips sonant. Frequencies up to 500 Hz are reproduced via a motional-feedback woofer system. Figure 2. The sonant. The electronic 'works’ are mounted on the inside of the hinged rear panel, which doubles as a heat sink. Figure 3. Amplitude-frequency response of the sonant. The solid line shows the response with motional feedback in operation; the dashed line shows what would happen rf there were no motional feedback. Figure 4. Distortion characteristic at nominal output. The solid line indicates the distortion with motional feedback operative. Removing the feedback results in the much higher dashed curve values. See the text for an interpretation of these curves! overdriven, when it suddenly starts to produce sharp-edged waveforms contain- ing musically-unpleasant high order com- ponents. Loudspeaker distortion at low frequencies, on the other hand, consists mainly of third harmonics. This low- order distortion merely disturbs the balance between fundamental and naturally-present harmonics in the in- | strument being reproduced, causing relatively acceptable ‘colouration’ of the ; sound. The curve in figure 4 can be viewed as . follows. Assume that the loudspeaker is operating without feedback and is being nominally fully driven with a pure 30 Hz tone. The delivered output will consist of 76% fundamental (30 Hz) and some 24% third harmonic (90 Hz). The loud- , speaker with feedback will (under equiv- alent conditions) produce 92% funda- mental and only about 8% third harmonic, so that the reproduction will sound much less ‘coloured’. At still higher (overdrive) levels the effect will become still more pronounced : operated without feedback thfe loudspeaker will produce as much as 80 to 90% distortion, reducing to about 30% with feedback operative. The operation of the feedback | is then to increase the amount of funda- mental produced from as low as 10% to I some 70% of the total. The great increase in the level of the fundamental inevitably makes the re- production more natural-sounding, more ‘realistic’. One then overhears remarks like: “It is as if there is no longer a loud- speaker getting in the way”. It is to be expected that this performance can only be obtained at a price. The price to be paid was already mentioned in an earlier article (‘electronic loudspeaker’ Elektor no. 1) — it is simply that the amount of sound output obtainable at ‘flat’ power-response is considerably lower in the feedback case. Maintaining the loudspeaker’s amplitude response flat below the point at which the basic system naturally rolls off implies the application of ‘brute force’ by the feed- back drive. Extending the response by an 3 13 dB 1* 0 -6 — 12 -18 -24 III ii hi IB i i IlHi ■ ii in iiaa 1 II III Hi ii hi Hi ■ i j|Hi m ■ ii in IlHi gj 1 i ii ■ m.f.b. on II III T I i 1 III m n ii i i a ii ill m m ii ii ii Hi ■ i i IIK^ ■ ii 1 ii; m in ii i h ii 9 il ii ii L i II 5 II 1 Hi ii ii \\m H ii 1 ii ii ii II S a ii ii r i ii II hi ii ii ii II ... I ii H 1 II a m ii ri a ii II 1618 3 1 0 20 50 100 200 500 IK 2 10* 20 50 100* 00 6 -fiHzi )0 1MHz octave would in fact require the ampli- fier rating to be increased more than 10 times. Any attempt to actually do this is of course to risk destruction of the driver — which in the Philips case is rated (as is the associated power ampli- fier) at 40 watts. The seemingly-obvious assumption that the power-response should be flat leads to the conclusion that the amplifier, at nominal output of the system above the natural rolloff fre- quency, always operates well below the level of which it is capable. Fortunately the assumption is incorrect! The continuous line in figure 5 is a con- tour for the maximum level encountered in ‘typical recordings’ as a function of the frequency. It was derived from measurements performed on a large number of disc records. The dashed line is a contour which applies to one or two extreme-case recordings (e.g. Decca’s ‘Zarathustra’). It may be pointed out that the extremes below 100 Hz are rarely encountered (‘Zarathustra’ or Saint-Saens ‘Organ Symphony’); but that the treble-range extremes are more common (e.g. percussion and synthesizer- effects in pop-recordings). This higher contour can be exceeded by 6 to lOdB during momentary signal peaks, mainly in the mid-range up to about 3 KHz. The right-hand vertical scale in figure 5 has been chosen to represent fairly loud music reproduction (as typically en- countered in monitoring rooms during classical recording). The dash-dot con- tour indicates (to this scale) the maxi- mum level of which the Philips sonant is capable in a fair-sized domestic listen- ing room. It will be clear that the system is capable of handling the ‘peak pro- gramme level’ discussed above at all fre- quencies higher than its 35 Hz amplitude- response rolloff point. The electronics The complete electrical circuit diagram of the sonant is given in figure 6. To improve the readability of this diagram it has been divided into sections. The sections A and B are the ‘woofer’ drive circuit and its feedback system; the mid-range and treble channel consists of sections C, D and E; power supply section G, finally, is controlled by signal- dependent shutdown F. The first part of section A (T421 to philips sonant T423) is an active bandpass filter with cutoff frequencies of approximately 35 Hz and 500 Hz. The motional feed- back signal is injected via C 506. The operating principles of such filters are (or should be) well enough known. The remainder of section A is a normal class B power amplifier, with an operating bandwidth of 5 Hz to 2 KHz and rated at 40 watts. It meets all the requirements of this application. One or two design de- tails may be worth noting: — the differential input pair T 424 and T 425, necessary to prevent ‘disagree- ment’ between the various feedback paths (C 506/R 603/T 423, R 61 1/ C 504 and R 623/C 509/R 622 C511/R619). — the extensive filtering in the above- mentioned feedback paths, which are designed to optimise the overall amplitude- and phase-characteristics of the system as a whole. — the application of power-Darlingtons in the output stage. — diode D 456, which ‘clamps’ the base- voltage of T 430 whenever this attempts to exceed the supply-rail voltage as a result of the ‘bootstrap- ping’ via C 513. In section B a dotted rectangle is shown enclosing the drive-unit itself, the accelerometer and an impedance- matching stage. The pickup device proper, plus an FET and two resistors, are mounted on a small PC board glued to the leading edge of the drive-coil former (figure 7). The pickup device is a small ceramic plate, suspended in an opening of the PC board by means of rubber blocks. The voltage delivered by the pickup is proportional to the force it experiences, which in turn (f = M.a) is proportional to the acceleration. The moving mass is in fact largely due to the solder droplets — so that these must be carefully controlled in size! Inevitable production tolerances can be corrected by means of preset potentiometer R 654. The ceramic plate performs best when looking into an extremely high im- pedance, which is the reason for using an FET. Installing this FET beside the pick- ■ up, rather than on the main PC board, avoids problems with hum and instabili- ty. The circuit around T 433 combines the functions of maintaining the FET at the correct operating point and of extracting the output signal for further processing, j The remainder of section B is a filter- amplifier. It delivers a signal strong enough for injection into the main channel, at the same time arranging for unconditional stability of the feedback system. The power amplifier circuit for the mid- I range and treble loudspeakers (section C) 1 closely resembles that in the woofer channel. In this case however the 500 Hz high-pass filter is built up around the first transistor of the amplifier itself T 439). The output stage is biased to a quite high value of standing current j (about 200mA), i.e. in class AB, to elim- I inate any possibility of crossover distor- philips sonant elektor february 1975 — 219 amplifier passes through the dividing network (section D) to drive the loud- speakers of section E. The circuit of section F is a kind of automatic supply switch. With mains applied and the on/off switch depressed there will be DC on the ‘+2’ and ‘+3’ supply rails. An input signal delivered to the sonant at a level higher than about ImV will, after being amplified and j rectified, cause the Schmitt trigger in I section F to change state and so pull down relay S 402. This turns on the feed I to the power amplifiers, within 1 second ! of the arrival of a signal. If the signal is interrupted for more than about 3 min- utes, the circuit assumes that the sonant is no longer required and shuts down the power amplifiers. The actual power supply circuit (sec- tion G) is fairly standard. Special atten- tion has been paid to the feed for critical circuits (‘+3’): T451/T452 provide extra smoothing by what is in fact gyrator-action. An additional effect of this arrangement is to cause the feed voltage to rise slowly after application of the mains, thus eliminating ‘thump’. The circuit including R 726, R 727 and D 471 is an interesting design-gimmick: the supply indicator lamp L414 is arranged to glow dimly so long as the sonant is dormant with mains ‘on’, via R 726 and R 727. When the amplifier feed is turned on, however, D471 effectively short-circuits R 726, so that the lamp glows at normal brightness. This provides, if nothing else, a ‘standby’ condition for the lamp! However, our most recent information indicates that this feature has been omitted from the latest models: the lamp switches on and off together with the amplifiers. M Figure 5. Contours of maximum level against frequency. The solid curve indicates the levels encountered in typical music recordings; the dashed curve shows higher levels occasionally encountered. The right-hand scale is chosen to represent fairly loud reproduction (classical monitoring); the dash-dot line shows the sonant's peak programme capability to the same scale. Figure 6. Complete circuit diagram of the sonant. 'A' and 'B' form the motional feed- back woofer-channel; 'C', 'D' and 'E' are sec- tions of the mid-range/treble channel; 'F' is the signal-sensing power-shutdown; 'G' is the power supply section. ® I Iffft/right *W(tCh (J> I input mode switch MFB loudsp«aker box 22RH532/00 220 — elektor february 1975 motional feedback motional feedback Many people are quite satisfied with the musical pleasure they can obtain from a well-designed reproducer. For the others, the pleasure is heightened by the intellectual satisfaction of knowing how the thing works and why it sounds the way it does. A brief outline of the theory of motional feedback is therefore in order. A good starting point is to take a look at the arrangement chosen forthe 22RH532 ‘Electronic’: a loudspeaker driver fitted with an accelerometer, mounted in a stiff airtight enclosure. This arrangement is shown in the block diagram of figure 1 . The acceleration of the cone is measured, the derived voltage being applied as negative feedback. A short calculation will show that this is the best approach. The radiated acoustic power (P a ) is given by: P a = u 2 • R a where u is the ‘particle velocity’ (equal to the cone velocity) and R a is the ‘radiation resistance’ (real part of the air-load on the moving cone). For frequencies below about 500 Hz the value of R a increases with the square of the frequency (fig- ure 2). The application of acceleration-depen- dent negative feedback will tend to keep the acceleration of the cone (a) linearly dependent on the input voltage (vj) and independent of frequency. The relation between acceleration (a) and velocity (u) of the cone is: , a v i u = a*t so that u cu co Figure 1. Block diagram of a motional feed- back system making use of an accelerometer. Figure 2. The radiation resistance (R a ) of the air-load on a typical moving-coil loudspeaker in box increases (up to about 500 Hz) with (jfi. Since it is the radiated power that is proportional to R a , the output at constant cone velocity would rise at 6dB/octave. Figure 3. Constant acceleration means a cone velocity that is inversely proportional to Cd This velocity is proportional to the square root of the radiated power. So u can be plotted as a — 3dB/octave slope. Figure 4. The radiated acoustical power is pro- portional to the product of radiation resistance and velocity squared. Combination of figure 2 with twice figure 3 yields a total slope of OdB/octave below 500 Hz i.e. flat frequency response! Since a is independent of to in this case, the velocity will be inversely pro- . portional to frequency (figure 3). This leads to: Pa»u a -R a = C-^*« a -v i a I where C is a constant. (If we ignore boxl dimension- and room position-effects!)! The relationship between radiated sound! power and input voltage is therefore inde- pendent of frequency (figure 4). In other, words, the amplitude-frequency response! characteristic is flat. Summary The amplitude-frequency responst characteristic (the ‘frequency response’] is determined by two terms: the radiation resistance and the cone velocity (squared). The radiation resistance rise) quadratically with frequency (up to about 500 Hz, figure 2). The velocity de creases in inverse proportion to the risin( frequency (assuming constant accelera tion, figure 3). The final result is obtained by combining figure 2 with twice figure 3 (velocity squared!) — which yields fig ure 4. H improving the readability of seven-segment displays elektor february 1975 221 improving the readability of seven-segment Seven-segment displays of various types are now the most popular format for digital display in many applications. The most common decoder-driver used with these displays is the 7446 (or 7447) which may be used with displays of the LED or Minitron type. The results obtained with these decoders are, in general, very good, but the format of the digits 6 and 9 leaves something to be desired. These digits are decoded as shown in figure lb and most people would agree that they are greatly im- proved by the addition of segments ‘a’ and ‘d’ respectively, as in figure lc. The Japanese tend to use this presentation in their electronic calculators. This format may be obtained with the 7446/7 by parallelling the ‘a’ and ‘d’ outputs with external transistors which are turned on when either a 6 or a 9 is displayed (see figure 2). The only prob- lem is to derive a suitable code from the BCD input to drive the transistors. A ‘1’ must be applied to the base of the appropriate transistor when either a 6 or a 9 is displayed. Looking at the truth table for the 7446/7 it is apparent that when a 6 is displayed columns B and C of the BCD input code are ‘I’. Column C cannot be used, how- ever, for looking at the rest of this column it can be seen that column C is also a T’ for digit 4. Since 4 does not utilise seg- ment ‘a’ input C cannot be used to drive the transistor for this segment. Input B may be used however since the other digits with a ‘1’ in this column are 2, 3 and 7 which all use segment ‘a’. Turning to digit 9 it can be seen that there are ‘IV in columns A and D of the BCD code. A obviously cannot be used since digit 1 also has a ‘1’ in this column but does not contain segment ‘d’. Column D may be used, however, since the only other digit with a T’ in this column is 8, which contains segment ‘d’ anyway. 14 Figure 1. a. Alphabetic designation of the seven segments of a display, b. Usual format of digits 6 and 9. c. Improved format of digits 6 and 9. Figure 2. The output stage of a 7446/7 seven- segment decoder with the external transistor in parallel. Figure 3. The complete circuit for the im- proved readability display using a 7447. The 7446 has an identical pinout. Truth table for the seven-segment decoder without the additional transistors. Digit D c B A a b c d e f 9 0 0 0 0 0 0 0 0 0 0 0 1 1 0 0 0 1 1 0 0 1 1 1 1 2 0 0 1 0 0 0 1 0 0 1 0 3 0 0 1 1 0 0 0 0 1 1 0 4 0 1 0 0 1 0 0 1 1 0 0 5 0 1 0 1 0 1 0 0 1 0 0 6 0 1 1 0 1 1 0 0 0 0 0 7 0 1 1 1 0 0 0 1 1 1 1 8 1 0 0 0 0 0 0 0 0 0 0 9 1 0 0 1 0 0 0 1 1 0 0 — * 1 a b c f L n r n e U 1 U J 1509- la 1509 -lb 1509 -1c 2 1 1509 -2 1509-3 222 — elektor february 1975 microdruit microdrum Linking up more or less with the minidrum published elsewhere in this issue of Elektor, this article describes the simplest member of the electronic drum family, to wit: the microdrum. In this micro version the number of instruments has been limited to two: one high and one (very) low bongo. Only two transistors are needed to make an instrument that can be connected directly to any power amplifier. Neglecting the cost, the obvious choice of instruments for the microdrum would be a bassdrum and a snaredrum. How- ever, the accurate imitation of a snare- drum requires quite a number of com- ponents because of the necessity for a noise generator, so that its application in a simple and, above all, inexpensive little “drumbox” is less attractive. For this reason, the instruments chosen for the microdrum described here are a bassdrum and a high bongo. In addition, to simplify matters even more, the sound of the bassdrum was lifted by one octave so that expensive bass loudspeakers are not required for reproduction. The result is a combination of a very low and a very high bongo. The circuit Figure 1 shows the circuit diagram of the microdrum. The bass sound is produced by means of the oscillator comprising T] (this is the low bongo sound); the oscillator built around T 2 represents the high bongo. Both oscillators are so de- signed that they are inoperative when the push button switches S] and S 2 are open. As can be seen in the diagram, the Figure 1. Circuit diagram of the microdrum. The sound levels of the bass and bongo can be adjusted to taste by experimenting with the values of Rg and R 2 O' oscillators are of identical design. If S 2 or S 2 is pressed, a positive pulse appears on the base of the corresponding transistor. The shape of this pulse is determined by a circuit consisting of Cj , R, and C 2 for the bass oscillator and R l2 andC 7 for the bongo oscillator. In contrast to the bass, the bongo has no parallel capacitor across Ri 2 ; hence the sound of this instrument is sharper. The discharge time, and thus the duration of the pulse, is determined by the values of R 5 and R 6 , and R 17 and R 16 respect- ively. The pitch of the bass oscillator de- pends on the values of C 3 , C 4 and C 5 ; the pitch of the bongo is governed by the values of C 8 , C 9 and C 10 . Where C 3 , C 4 and C 3 are concerned, the best values found to lie between lOnF and 47nF; the best values for C 8 , C 9 , and Ci 0 are be- tween 4n7 and lOnF. Of course, everybody is quite free to ex- periment with other capacitor values. A higher value for the capacitors gives a lower tone. The signals are taken from the collectors of the two transistors and fed to the out- put via capacitors (C 6 ,C n ) and resistors (R 9 ,R 20 ). The sound levels of the bass (i.e. low bongo) and the high bongo can be adjusted to taste by experimenting with the values of R 9 and R 20 . The rms output voltage of the microdrum is about 1 V; the minimum load impe- dance is 1 8k, so the combination can be connected to almost any power ampli- fier. The supply voltage can lie between about 1 5 V and 24 V. If mains supply is used, it need not be stabilized. As the total current consumption is no more than about 5mA, the apparatus also per- forms well when fed from torch batteries. Even if used intensively, these batteries will last for months. Figure 2 shows the printed circuit, and figure 3 shows the component arrange- ment. The modest dimensions (abt 7.5 X 5 cm) leave ample possibilities in choos- ing a suitable cabinet. Additions As already explained, the sound of the instruments can be changed by using other values for the capacitors C 3 , C 4 and C s (for the bass), and for C 8 , C 9 and C 10 (for the high bongo). It will be clear that further experimentation can lead to en- tirely different instruments whilst using the same oscillators. A few short tests carried out in the Parts list Resistors R 1,R 1 2= 27k R 2- R 3- R 10- R 11> R 13. R 14 = 10k r 4 - r 1 5 = 1 00 k r 5, r 17 = 560k R 6- R 16 = 270k r 7> r 18 = Ik r 8< r 19 = 2k7 Rg, R 20 ' 1 5k (see text) Capacitors Ci.Cn =0.1 H C 2 .C 6 .C 7 = 0.47 JU C 3 .C 4 .C 5 = lOn (see text) Cg.Cg.Cio = 4n7 (see text) semiconductors T 1 .T 2 = TUN Miscellaneous Si,S 2 = single pole push button switch microdrum Elektor laboratories have indicated that changing the values of C 8 , C 9 and Ci 0 to 2n7, produces a sound rather Like wooden blocks. It might therefore be attractive to ex- tend the microdrum with one or more identical p.c. boards, each with different capacitors. universal display elektor february 1975 — 223 universal Figure 2. A simple, yet complete electronic musical instrument on a printed circuit board of no more than 7.5 X 5 cm! Figure 3. Component arrangement on the board of figure 2. display It is frequently necessary to have available a numeric display for many projects such as frequency counters, digital voltmeters etc. It is a tedious and untidy business to build up such displays on matrix board, so Elektor have designed a universal display which should satisfy the requirements of most enthusiasts. The display may be used with seven-segment indicators of the LED or Minitron type. Construction Double-sided boards are employed in the construction of the display module and it may be seen from figures 6 and 7 that components are mounted on both sides of the board. It should be em- phasised here that great care is required in the assembly of these boards due to the degree of miniaturisation involved. The soldering iron must have an extremely fine tip and soldering must be done extremely quickly to avoid peeling the fine track from the board. The boards available from Elektor employ plated- through holes, so that it is unnecessary to solder to component leads on both sides of the board. Simply solder on the opposite side of the board to that on which the component is mounted. The counter/latch module The counter/latch module increases the The universal display is modular in con- struction and its basic form consists of a board to accomodate two displays and their associated decoders. The system may be extended to any number of digits and decade counter/latch boards may also be added. The universal display uses the popular 7447 decoder. The display format of this decoder is given in figure 1. However, for digits 6 and 9 the improved format described elsewhere in this issue is em- ployed. This is shown in figure 2. The basic configuration of the decoder with the additional transistors is shown in figure 3 and the complete circuit of a display module for use with LED displays is given in figure 4. That of the Minitron version is shown in figure 5, the only difference being that the Minitron does not require current-limiting resistors in series with each segment. Figure 1. The display format produced by the 7447 decoder. Figure 2. The improved presentation of the digits 6 and 9 as used in the universal display. Figure 3. The circuit used with the 7447 to achieve the improved 6 and 9 display. 224 — elektor february 1975 universal displi I 1 : : ' rr~at — t Components list for figure 4: Resistors: R 1 - R 2 - R 10- R 11 = 2k2 R 3 to Rg, Ri 2 to Ri 8 = 1 80 S2 Semiconductors: T| to T 4 = TUN IC 1 ,IC 2 = 7447 Li,L 2 = e.g. H. P.5082-7730 or 7750, OpcoaSLA I.T.I. Til 302, Data Lit DL707. AFC E IC2 DC B © B C 131 15 | 14 | 9 | 10 | 11 | 12 1 A F G E D C B IC1 gnd 7447 ( a e c d 'I 'I 2| 6| versatility of the universal display to a large extent. The circuit of the module is given in figure 8 and consists of two cascaded 7490 decade counters and two 7475 latches. The operation of these devices will not be discussed in detail as they are extremely commonplace; suffice it to note the following points: — the latch is enabled by a ‘1’ on the clock input. — the counter counts on a negative- going edge. — the counter is reset by a ‘1’ on the reset input. — for reliable operation of the counter the slope of the pulse edges should be greater than 2 V/p sec. Assembly of the complete display The board and layout for the counter/ universal display elektor february 1975 — 225 —05V Components list for figure 5 Resistors: R-| to R 4 = 2k2 Semiconductors: T 1 to T 4 = TUN IC 1 ,IC 2 = 7447 Display: Mi.M 2 = Minitron 3015 F _ IC2 © 7447 Figure 4. Circuit of the LED version of the universal display. Note that the decimal point series resistor has a higher value than the segment resistors to achieve the same luminous intensity. Figure 5. The Minitron version of the display which does not require series resistors for the segments. Note that the Minitron has a right- decimal point whilst the LED displays used have a left-hand decimal point. Figure 6. The p.c. board and layout for the LED display. The track shown in feint in the component layouts is the side of the board on which the components are mounted, i.e. the components are mounted directly on top of the track shown. Figure 7. The board and component layout for the Minitron display. The same remarks apply as for the LED display. 226 — elektor february 1975 Figure 8. Circuit of a two-decade counter/ latch. The necessary interconnections are of course made on the p.c. board. When the reset input is at '0' the counter will count. When input E is at 'O' the latch will store the information present on its inputs at the time the transition occurred. Figure 9. The circuit board and layout for the counter/latch. The same remarks apply as for figures 6 and 7. Note the connection slots on the right-hand edge of the boards for supply, reset and latch-enable busbars. Figure 10. A cut-away drawing of the com- plete universal display module showing the mounting of the counter/latch board and the busbars to other decades. The mounting of the counter latch board must be done accurately if several modules are to be cascaded. latch module are given in figure 9. The board is double-sided and the same constructional points apply as for the display module. The assembly of the complete universal display requires some care. The counter/ latch board is mounted perpendicular to the back of the display board. The BCD outputs a, b, c, d, A, B, C, D ,and the supply connections on the edge of the counter/latch board mate up with the corresponding connections down the middle of the display board. The method of construction is as follows: sufficient right-angle links are made from stout copper wire and soldered to the counter/ latch board so that they stick out parallel to the plane of the board but perpendicu- lar to the edge. The links are then pushed through the back of the display board so that the edge of the counter/latch board is flush with it and are then soldered (see figure 10). Interconnection of several boards Any number of modules may be easily interconnected to form a decade counter of any desired length. The modules are first joined mechanically by using 6 B.A. spacers and short lengths of 6 B.A. studding through the hole in the corner of the counter/latch boards. All the common interconnections between the boards, i.e. supply connections, counter reset and latch enable are made simply by running a wire bus in the slots on the top edge of the counter/latch boards. The counters are cascaded by connecting a wire link from the output of one stage to the input of the succeeding stage again on the top edge of the counter latch boards (remember to connect the output of a board to the input of the board to its left). The photograph should make this clear. M dil led probe elektor february 1975 — 227 led probe The use of digital ICs has increased over the years to such an extent that by now on average four to five ICs are used in a relatively small circuit. This does not make trouble shooting any easier; for this reason a universal tester was designed which can be used for testing ICs under operating conditions. The probe discussed in this article is suitable for testing 14- and 16-pin dual-in-line ICs. Although in principle the probe is suit- able for TTL- DTL- and cos/mos ICs, it is not recommended for use with the latter. In the first place the supply voltage for cos/mos ICs covers too wide a range (be- tween 3 V and 1 5 V). In the second place many cos/mos ICs can only deliver (or sink) a very small load current, so that the digital IC-probe may give erratic indica- tions. If the probe is to be used universally for DTL- and TTL-ICs, several requirements will have to be met. For ease of use, the probe must be able to function without an extra power supply. TTL- and DTL-ICs have a supply voltage between about 4.75 V and 5.25 V — apart from a few exceptions. This supply voltage for the IC under test must also provide the supply for the probe itself. The complete probe should draw as little power as possible from the IC under test, in order not to disturb the functioning of the circuit. In the design this has been achieved by using buffer ICs which ensure that the IC under test is hardly loaded, apart from the supply. The best way to obtain a simple test procedure is to have two rows of LEDs on the probe, each one corresponding to one pin of the IC under test. Each LED will light up if the corresponding pin is at the ‘1’ level. Consequently, a dark LED indicates that the pin is at ‘0’ level. The supply Certain problems arise from the require- ment that the probe must be fed from the IC under test. The supply is not always at the same pin; for 14-pin ICs and 16-pin versions there is a certain amount of standardisation, but with so many excep- tions that the probe must be able to locate the supply connections on any of the 14 or 16 pins. In its simplest form a ‘supply finder’ can be made as shown in figure 1. If in this figure point X has a positive voltage, this voltage will also be available on point A. A zero potential at point X will be passed on to point B. Because there can be up to 16 IC pins from which the supply must be found. Figure 1. A single ‘supply finder'. Figure 2. A supply locator with three inputs. Figure 3. A single supply locator with corre- sponding LED indicator. Figure 4. A complete supply locator with inverter and LED, of which 16 are needed. V. ■ L’.DUG-> 1 r wi D2 1 Hdug oB 5027 1 2 02 D3 z v n 04 1-4 1^ 05 * z n 6xC 06 >UG ^ 5027 : 3 . £!iPUG . 5 W\ Pi. DUG 1“ v R 1 >\D3 1 68001 l«< 027 3 LED 4 V- . Sl.DUG . rn Pi.DUG D3^ ^>0 -f330S>}-J 502 7 4 LED the circuit of figure 1 must be repeated 16 times. To clarify matters, figure 2 shows a ‘supply locator’ with three connections (X, Y and Z). If on one of these connec- tions, say X, there is a positive voltage, whereas point Y is at zero potential, the given supply voltage between X and Y is also available at points A and B. Because of the diodes, the voltage between the points A and B is twice the diode threshold lower than the voltage between X and Y. To keep this voltage drop at a minimum, it is better to use germanium diodes (DUG) than silicon types (DUS). Since a ‘ 1’ is always somewhat lower than the supply voltage (and a ‘0’ higher than the ‘actual zero’), a logic level will hardly contribute to the supply current for the probe. This implies (figure 2) that a logic level at point Z is scarcely loaded by the supply. Read out The logic levels are read out with LEDs. Figure 3 shows how this could be done. TTL- and DTL-ICs can handle the highest currents at an output ‘O’. It is therefore recommended to have the LED light up at a ‘O’. To keep the overall power con- sumption of the probe within reasonable limits, the LED (D 3 )of figure 3 draws no more than about 4 mA. This is sufficient to obtain a good optical indication. How- ever, 4 m A is too high a load for some IC outputs. Furthermore, the abovemen- tioned ‘0’ indication is less atrractlve. Both problems can be solved by adding an inverter to the circuit of figure 3. Fig- ure 4 shows a complete indicator stage, 16 of which are used. In this figure, I is the inverter mentioned. The result of the circuit according to figure 4 is a lower IC loading (about 1.5 mA per connecting pin) and a LED indication at ‘T. Owing to the TTL/DTL properties, in- verter I (figure 4) also ensures that the LED regards an unconnected pin as a ‘1’. This is the case, for instance, if a 14-pin IC is tested: the two remaining vacant pins cause their LEDs to light up. 228 - elektor february 1975 dil led probe XJUUUUUU ^ # ^ jW W" ^ Tunable aerial amplifier There is a mistake in the component layout for the PC board (elektor 1 , p. 42). One tap of coil Lj is shown connected to ground, whereas it should be connected to the junction of R 2 and C 2 as shown in the circuit diagram (figure 2), which is the neighbouring pin on the coil former. The PC board itself is correct. The FETs mentioned in this article are supplied by the following manufacturers: Siliconix: E300, E310, U310, E304; Teledyne: U1994E, 2N4416, 2N5397; Signetics: SD201. -FIDO, a new electronic game -dynamic range compressor -LED level indicator -time-lapse cinematography -TUP/TUN tester -touch control preamp -high quality disc preamp for dynamic cartridges The complete tester Figure 5 shows the circuit diagram of the complete digital probe. Some 32 diodes (D, ... D^) are needed for the voltage . probe. For the 16 inverters use is made of three TTL-ICs, type 7405. Each 7405 contains six inverters, so that two in- verters remain unused. The encircled codings in figure 5 represent the connect- ing pins of the IC probe. Points A and B are the supply and common for the three 7405 ICs. The printed circuit boards Figure 6 gives the lay-out of the two p.c. boards for the circuit of figure 5. The component arrangements for this circuit are shown in figure 7. Board A carries * only one 1C. This board accommodates the circuit section for pins 1 ... 8. After assembly the boards are connected back- t, to-back. This is illustrated in figure 8, showing a photograph of the complete probe. As the figure shows, the p.c. boards are placed with the component sides facing each other. A 16-pin con- nector is provided between them. The so-called pick-a-back connector shown is made by Electrosil. As shown in the figure, the anodes of the LEDs are coupled together and con- nected to points G and G respectively. The points A ... F on board A are electric- ally connected with the corresponding points on board B (so: A with A , B with B' ...). If this is done with stiff copper wire, the sandwiched assembly will be fairly sturdy. The opposing corner holes of the p.c. boards are also joined by means of rigid copper wire. H Zke missing link Modifications to Additions to Improvements on Corrections in Circuits published in Elektor elektor february 1975 — 229 Figure 5. The complete circuit diagram of a digital IC probe. Figure 6. The lay-outs of the printed circuit boards A and B for the probe. Figure 7. The boards A and B with the com- ponent arrangement for the circuit of figure 5. Figure 8. Photograph of a complete digiprobe. be dil led probe 230 — elektor february 1975 elektor 'sonant' loudspeaker system elektor „sonant loudspeaker system An earlier article explained the operating principle of the electronic loudspeaker. It was shown that this approach enables a woofer in a relatively small en- closure to provide really good reproduction quality from 40 Hz upwards. The second step is to describe a complete system, using active crossover networks, based upon the electronic loudspeaker used as a woofer. The system provides unusually good reproduction quality for its modest size and low cost. An electronic woofer can be designed to have a flat frequency response curve be- ginning at 40 Hz — or an even lower fre- quency if desired. This flat response will normally be maintained up to about 400 Hz. Above this frequency it will be necessary to apply additional correc- tions, which are dependent on the loud- speaker used, the size and shape of the enclosure (and the loudspeaker’s position on the front) — and even the position of the enclosure relative to room bound- aries. The simplest design choice would there- fore seem to be an application of the electronic loudspeaker to the woofer- range, with reproduction of the mid- and treble ranges by normal units. Such an approach makes it very desirable to use a separate driving amplifier for the woofer, with crossover networks ahead of this amplifier. The mid- and treble ranges can then be handled by one or two smaller amplifiers, for two- or three-way systems respectively. Two-way system The so-called two-way system provides a simple solution which can nonetheless produce excellent results. The block dia- gram for this is shown in figure 1. The ‘crossover’ frequency chosen is 340 Hz. The electronic woofer reproduces the range 40-340 Hz, while a separate 1 0-watt amplifier takes care of the remainder of the frequency range. Two preset potentiometers, inserted in the circuit ahead of the crossover net- works, enable the correct loudness bal- ance to be set up. The adjustment is best done by ear, preferably with the system installed in its working location. In this way the inevitable differences between the efficiencies of the various loudspeak- er-drivers used can be equalised. Care must be taken to connect the two drivers in the correct phase-relationship. The woofer is operated with the ‘plus’ terminal — the terminal which, when driven positive, causes the cone to move outwards — connected to the ‘hot’ power- amplifier terminal. The mid -high-range driver is then connected with its ‘plus’ terminal to chassis, i.e. the ‘wrong way around’. On many loudspeakers the ‘plus’ terminal is identified by the maker, by means, for example, of a red dot. Three-way system The three-way system is built up accord- ing to the block diagram given in figure 2. The electronic woofer operates as before, for frequencies up to 340 Hz. A second amplifier handles the range 340 to 4800 Hz., while a third unit operates at frequencies above 4800 Hz. This may seem to be the expensive way of doing things; but in practice the additional lower-power amplifiers often turn out to be less expensive than conventional cross- over networks. In any case the results are usually better, while the level-balance between the high- mid- and bass-loud- speakers can be very conveniently obtained by adjustment of the three pre- set potentiometers. Once again, attention must be paid to correct relative phasing of the drive-units. This is also shown in the block diagram. The crossover networks The crossover networks for the two-way system are built up according to figure 3; the three-way circuit diagram is given in figure 4. The circuits are simple, to the point, of being primitive, being cascades of passive RC-networks, buffered by means of emitter-followers. Reliable, stable and predictable. Furthermore the two arrangements are so similar that both can use the same printed circuit board. For the two-way system all that is re- quired is a single printed circuit board, with the component layout shown in figure 6. The three-way system circuit requires two of the boards. One of these is laid out according to figure 7, for the mid-range and treble channels. The other is laid out according to figure 8, with the circuits for bass reproduction via the electronic loud- speaker. This latter p.c. board also in- cludes an innut huffer tT, 1. elektor 'sonant' loudspeaker system elektor february 1975 — 231 Figure 1. Block diagram of the two-way system. high-pass (f 0 * 340 Hz) (12 dB/oct.) Figure 2. Block diagram of the three-way system. Figure 3. Circuit diagram of the two-way system. The motional feedback in the bass- channel is adjusted by means of P 3 ; the level- adjustment presets Pi and P 2 enable the correct loudness-balance to be obtained between bass and treble channels. low-pass filter 1 (fo * 340 Hz) (12 dB/oct.) low-pass filter2 (fo * 40 Hz) (6 dB/oct.) electronic loudspeaker mid-range band-pass (12 dB/oct.) low-pass filterl (f 0 * 340 Hz) (12 dB/oct.) low-pass filter2 (fo * 40 Hz) (6 dB/oct.) treble CVA/ ^ 1 ■ P"J V mid-range iow P"J 20Wj>— ■ bass 4-r 1 electronic loudspeaker Parts list for figures 3 and 6 Resistors: R 1 = 220 K R 2- R 6 = 100K r 3' R 5- r 1 1 • R 1 3 = IK R 4* R 9' R 12' R 16 = 6K8 r 7 = 56 K r 8 = 3K3 R 10 = 4K7 r 14- r 18 = 22K R 1 5 = 2K2 r 17 = 1K8 r 19 = 1 S2 P1.P2 = 5K (preset) P3 = 1 K (preset) Capacitors: Cl = 6n8 c 2- c 5 = 470n c 3 - 330n C 4 = lOOn c 6 = 180n c 7 = 10/U/16V Transistors Ti to T 4 : TUN R2 47 o„ r J -|R5 2x 340Hz H 20W > 2 x340Hz 40/ 400 Hz 1527 - 3 232 — elektor february 1975 elektor 'sonant' loudspeaker systevr Parts list for figures 4, 7 and 8 Resistors: Irioi nmos R 1. R 101> R 106 r 2- r 6- r 102- r 107 r 3> r 4- r 9- r 12- r 16- r 104' R 109- r 110' r 112> r 113' R 116 r 7 R 8> R 108' R 111 R 10 R 1 1' R 13< r 103- r 105, r 1 1 5< r 1 18 R 14' R 18 r 15 r 17 r 19 Pi p 2- p 101- p 102 p 3 = 220k = 100k = 6k8 = 56k = 3k3 = 4k7 = Ik = 22k = 2k2 = 1k8 = i n = 50k (preset) = 5k (preset) = Ik (preset) Capacitors: C-|,C 3 = 330n C 2 .C 7 = 10/J/16Vr\ C 4 = lOOn V C5.C107 = 470n Cg = 180n Cl 01 = 470p C102 = 33n Ci 03 = 6n8 Cl04- C 105 = 4n7 TR2 R4rh ’OM : eo 16V US Transistors: Ti to T4, Tioi to T-|04 = TUN PI = total level CIO! II L, 6 ll n PI 01 = treble level PI 02 = mid-range level P2 = bass level &r ° 2 oc •n 33n X R104 - sL_ y. 2 x 4.8kHz piRin n Rm ^108 3 N T103H- TioaJ-r 1 > TUN /'7/\tun /TV UN 13 - 2x 4.8kHz Rl6f| I0p ® 16V 40/400 Hz Setting-up procedure The adjustment of the electronic woofer — in either arrangement - is made by means of the preset potentiometer P 3 . This is slowly turned up from zero posi- tion to the point at which the system starts to oscillate (howl), and then turned back until oscillation just ceases. Some loudspeakers have inferior mag- netic systems in which the degree of electro-magnetic coupling varies during the drive-coil throw. An adjustment made as above can then give rise to a kind of ‘after-pong’ effect (if once you hear it you’ll know what we mean!). This can be clearly demonstrated with a square- wave input; but of course it can be objec- tionable on some kinds of music pro- gramme. The remedy is very simple: back off a little more on preset P 3 . In the two-way system, the balance of loudness between the bass and treble channels is adjusted by means of presets P, and P 2 . The same adjustment in the three-way system is made with P, 01 , P^ and P 2 . Turn the preset which controls the least sensitive channel to maximum, then adjust the other(s) until the balance ‘sounds right’. It is well worthwhile spending a little time on this. Choice of components The components from which this system is built up — the amplifiers, drivers and the enclosure — have to meet certain specific requirements. The amplifier used with the electronic woofer must be completely and uncon- ditionally stable, even when its output is short-circuited. The prototype systems scribed (Elektor no. 1). The amplifiers used in the treble channel(s) may be small high-quality units. They are not normally called upon to deliver as much power as the woofer’s drive-amplifier, but they must of course be free from audible distortion. The bass driver should preferably have a high-quality electro- magnetic motor system. This ultimately determines the system performance and the maximum obtainable output. It is also desirable that the cone-suspension be moderately stiff; the supercompliant rubber surrounds on some woofers can misbehave quite seriously at high drive- levels in a small enclosure. One will then need a larger — possibly damped — en- closure. During measurements on various makes and types of bass drivers it became clear that the Philips 97 1 0 (M) is a particularly suitable unit. The same maker’s AD 3701 also did well in the tests, but it has in the meantime been replaced by the nominal- ly almost identical (according to Philips) AD 7061 M which we have not yet ex- tensively tested. The treble reproduction in these systems is ‘standard’ — without use of the elec- tronic loudspeaker principle — so that the dirvers have to meet normal hifi-require- ments. Among the several units that seemed to give good results are the mid- range drivers Kef B 110 and the Philips AD 5060/Sq, along with the ‘dome’ tweeters Kef T 27 and Heco PCI! 24. The woofer enclosure can be a fairly simple design. What is required is a totally closed box having fairly solid walls — we suggest chipboard of 15 or 18 mm thick- ness. The volume is not critical since it really only affects the low-end power- response. For typical domestic listening 15 litres is usually sufficient (e.g. 12” X 9" X 8"). A little damping is desirable, particularly if the box is made consider- elektor ‘sonant* loudspeaker system elektor february 1975 — 233 Figure 4. Circuit diagram of the three-way system. P 3 is once again used to correctly set up the bass-channel motional feedback; loud- ness-balancing is done by adjusting P 2 , P -|01 and P' 102 - Pi provides an additional total-level adjustment, which can be convenient when set- ting up a stereo pair. Figure 5. The universal printed circuit layout for all filter circuits. Figure 6 . Components layout for the single p.c. board used with the two-way system according to figure 3. Figure 7. Component layout for the mid-range and tweeter channels of the three-way system (figure 4). Figure 8 . Component layout for the bass- channel and buffer stage as used in the circuit of figure 4. Figure 9. Result of measurements (total radiated power) on a typical three-way system. ably larger than 1 5 litres. This will usually arise if one wants an extended power- response. Damping may in any case be needed with drivers having limited mag- netflux (to help eliminate the ‘after-pong’ effect). A 20 mm thick pad of glasswool, mounted by means of laths at a random angle through the centre of the enclosed volume — not parallel or close to the walls — will do the trick. Conclusion Application of the electronic loudspeak- er-compensation described here — actual- ly a form of ‘motional feedback’ — can enable excellent results to be obtained with a relatively small woofer-enclosure. The curves in figure 9 are the results of measurements made on a typical system, with and without the compensation operative. The only real objection to this approach is the fact that it requires fairly critical adjustment for best results. This difficul- ty would be considerably lessened if a bass-driver fitted with a properly- designed and reliable feedback- transducer were to become generally available. The transducer could deliver a signal proportional to the cone’s dis- placement, to its velocity, or — preferably — to its acceleration. One recently- introduced commercial system (Philips) is designed around just such a woofer- accelerometer combination. As a final remark we note that a loud- speaker operating with motional feed- back forms part of a control-loop which includes the power amplifier. It is ob- viously desirable to install this amplifier in the loudspeaker cabinet, or the fact that the feedback signal is carried by the same leads which deliver the drive power means that the system will require re- adjustment every time the lead length is changed ! H r 234 — elektor february 1975 tv sound t #" ^ I I W 1 W ^ For those who are dissatisfied I I with the quality of their ^ television sound (and who isn't here is a design which enables a high-quality sound signal to be derived from a television receiver and fed into a separate amplifier and speaker system without making any connection to the T.V. set; a procedure which can be highly dangerous in view of the live-chassis construction employed in the manufacture of television receivers. Television Sound, bad or indiffer- ent, but hardly ever good Many people possess audio equipment of at least moderate quality from which they expect to obtain good sound with discs, tape or F.M. tuners, but even audiophiles frequently put up with mediocre noises from the ‘idiot’s lantern’. This is a pity as T.V. producers nowadays tend to give much more thought to the artistic quality and suitability of music which is part of a T.V. programme. In fact, even the ‘leitmotive’ to some T.V. series are so popular that they are released on records. People who appreciate high quality sound but who do not have a technical background (this includes many news- paper programme critics) tend to place the blame for poor quality sound on the shoulders of the BBC or IBA. This is most unfair as the quality of the sound leaving the studio is generally fairly high. The blame lies with the manufacturers of T.V. sets who skimp on the audio side of their products in the interests of economy and compactness. They can get away with this because the ear will tolerate poor sound quality much more than the eye will tolerate defects in the picture. The audio amplifiers in T.V. sets are generally of low power and poor performance. In addition it is difficult to put a good-quality loudspeaker into a television cabinet because a) there is insufficient space (except in large floor- standing models) and b) good loud- speakers generally have powerful mag- nets which tend to upset television tubes, especially colour ones. One solution to the problem of poor T. V. sound would be to have a separate U. H.F. tuner coupled to a Hi-Fi system, and there are such tuners on the market. This would, however, duplicate a part of the T.V. set which does its job perfectly well (bad T.V. sound invariably starts after the detector) and in addition retuning would be necessary whenever the T.V. channel was changed. Another possibility would be to extract the audio signal after the detector via an isolating transformer, but this involves tinkering with the T.V. circuitry, which is not always possible, expecially if the set is rented. The circuit to be described avoids all these difficulties by extracting the 6 MHz intercarrier sound signal by means of a pickup coil on the back of the set so that no electrical connection is required. The circuit picks up the sound signal for whatever channel the T.V. is tuned to, so that no retuning is required when the channel is changed. Principle of Operation In figure 1 the signal from the pickup coil is amplified and filtered with a passband of 300 kHz. This filtered inter- carrier signal is fed to one of the inputs of the phase comparator ICj . The other input comes from a voltage-controlled oscillator (VCO). The output voltage of the phase comparator is proportional to the phase difference between its two inputs. This output is fed through a low- pass filter to the control input of the VCO. This control voltage alters the VCO frequency so that it tends to to become the same as that of the intercarrier signal. When a VCO is ‘locked-in’ to a fixed-frequency signal the output of the low-pass filter is constant. However, the frequency of the intercarrier signal is not constant as it is frequency-modulated and the VCO frequency must follow the changes in frequency to remain locked-in. This means that the VCO control voltage must change. Since the change in control voltage is proportional to the change in frequency it follows that the changes Table Performance Data Signal-to-noise ratio (Audio signal pro- duced by 1 60 /JV intercarrier signal with ±40 kHz deviation at 1 kHz versus R.M.S. noise level with no input) - greater than 52 dB A.M. rejection with 160pV intercarrier signal 85% ampli- tude-modulated - greater than 40 dB PLL capture range -1 MHz PLL hold range -4 MHz Supply voltage - 1 2 V ± 2 V Supply current - 20 mA elektor february 1975 - 235 Figure 1. Block diagram of the television sound unit. The phase comparator, low pass filter and VCO form a phase-locked loop (PLL) F.M. detector. Figure 2. Complete circuit of the television sound unit. in control voltage are the audio signal which modulated the intercarrier signal. This phase-locked loop system thus demodulates the 6 MHz intercarrier signal and all that remains is to de-emphasise and amplify it. Circuit Details The complete circuit, including the power supply, is mounted on a single printed circuit board. The only external connections (apart from the mains lead) are the connections to the pickup coil and the audio output. Figure 2 shows the complete circuit. 236 — elektor february 1975 tv-sound The pickup coil leads are connected via isolating capacitors Ci and C 2 to the inputs of the differential amplifier Ti and T 2 . This means that common-mode signals such as mains are largely elim- inated and only signals induced in the coil will be amplified. Diodes Di and D 2 clip any high-voltage transients which might damage T 1 and T 2 . T 3 , D 3 and D 4 form a constant-current source for the differential pair. The ceramic filters Fn and Fj 2 each have a passband of 300 kHz between the 3 dB points, centred on 6 MHz, so that two in cascade, with T 4 as a buffer between them have a 6 dB passband of 300 kHz. Ds , D 6 and C 7 form a peak- detector which rectifies the signal on the output of F 12 to drive a 150/iA meter to indicate the signal strength. Pi adjusts the sensitivity of this meter. The signal at the output of F 12 is further amplified by T s and is then clamped _ by D 7 and Dg. Since the intercarrier | signal is derived by mixing of the sound and vision signals there is a very high level of superimposed A.M. due to the video modulation. For this reason the F.M. demodulation process must have - very good A.M. rejection and so a phase- - locked loop (PLL) system was chosen. The CA 3080 (lCi) is described by the manufacturers as an operational trans- : ^ Components List: Resistors: R 1. R 2. R 6. R 36= 100ft r 3- r 5. r 7- r 11 = 10 k r 4- r 9' R 13> r 15< r 16> r 21 > r 23 to r 26> r 37 = 1 k r 18- r 28< r 29. r 35 = 100 k r 30- r 31 = 15 k R 8 .R 12 = 2k2 r 10- r 14 = 230 R-17 =47 k Rl 9 = 560 f2 R 2 o = 220 k R 2 2 = 68 k R 2 7 = 4k7 r 32 = 5k6 r 33 = 1 k 2 FI 34 = 470 n Semiconductors: ICi = CA 3080 (RCA) T-| to Tg = BF 199 Ti 0 = BC 1 08 Tn = BC 177 A T 12 = BC547 D-i to D 10 = 1N4148 Du = 12 V/400 mW zener diode B = 4 X 1 N4002 Miscellaneous Tr = Transformer 12 V/50 mA FLi,FL 2 = 6 MHz ceramic filters, e.g. SFE 6 MA (Murata) tv-sound elektor february 1975 — 237 conductance amplifier (try saying that after a few pints) and is used here as an asymmetrical phase comparator. Its significant feature is that the gain can be changed by altering the current into pin 5, the relationship being linear. T 6 and T7 form an astable multivibrator. Diodes D9 and Djo limit the collector swing of these transistors to about 0,7 V so the multivibrator will run at higher frequencies than it could if larger swings were allowed. The running frequency of the multivibrator can be controlled by the current into the commoned bases of Ts and T9. (So this is really a current controlled oscillator but we call it a VCO for consistency.) Coarse and fine preset adjustment of frequency is provided by P 2 and P3 respectively in the common emitter bias circuit of Ts and T9. A current feed into pin 5 of ICi is derived from the collector of T 6 . The clipped intercarrier signal is fed into one of the differential inputs of ICj (pin 3) whilst the other input is grounded to signal via Cn- Since the gain of IC t is proportional to the output voltage of the VCO it is clear that the inter- carrier signal and the VCO are being multiplied together and the output current of ICi varies according to the relative phase of these two inputs. The output current of the IC is taken through a low-pass network and this filtered output controls the frequency of the VCO by injecting current into the bases of T g and T9. The sense of the control current is such that it always tries to keep the VCO in phase with the inter- carrier signal. The control current there- fore follows the variations in frequency of the intercarrier signal due to the audio modulation. The audio signal which thus appears on the control input of the VCO is de-emphasised by R 2 7 and C19 and is amplified by Tjo and Tu- The simple power supply circuit is also shown in figure 2. It consists simply of a transformer, bridge rectifier and smoothing capacitor and an emitter- follower stabiliser with zener diode D n as voltage reference. C24 suppresses noise from Dn . Construction and Alignment The construction of the p.c. board is straightforward and requires no further comment. The construction of the pickup coil is shown in figure 3. The dimensions given should be adhered to as the coil inductance and lead capacitance are designed to be broadly resonant around 6 MHz. Single-strand 22 SWG plastic-covered wire should be used for the coil and the twin con- necting lead should be made up of two lengths of screened audio lead with an outside diameter of 3 mm. and poly- thene dielectric. The leads are shown separated in a) to make the connections to the screens clear, but thay should actually be touching, as in b). To align the circuit, connect the T.V. sound unit to the mains and check that all the D.C. voltage levels are correct. Connect up the pickup coil and connect the audio output to a suitable amplifier. Turn the slider of P! to maximum. Tune the T.V. set to a convenient programme and search over the back of the set with the pickup coil until the meter shows a deflection and position the coil for maximum deflection. If the meter goes off the scale turn down Pi . Adjust P 2 and P 3 until the VCO locks in and a sound signal can be heard. If no sound results it is possible that unwanted pickup is coming from the line output transformer, so reposition the pickup coil and try again. When a sound signal has been obtained move the pickup coil around until the background noise is a minimum. This is not necessarily the position which gives maximum deflec- tion on the meter. It now remains to set the VCO free-running frequency to 6 MHz so that it is near the middle of its ‘capture’ range. That is to say the frequency at which the VCO tuns when it is not receiving an intercarrier signal should be around 6 MHz. If it is too far away from this frequency the system may fail to lock-in when next switched on without being adjusted, since at the first trial it may have been within its ‘hold’ range, but outside its ‘capture’ range, which is narrower. 3 Proceed as follows. Having found the optimum pickup coil position on the back of the T.V. set move the coil •away from the set until the signal disappears. Move the coil back towards the set and check that the sound reappears. It should be possible to adjust P 2 and P3 so that the signal reappears whilst the coil is still a few cm. from the back of the set and this is satisfactory. One fault which the T.V. sound unit will not cure is ‘buzz’ due to captions, subtitles and other high contrast areas of the picture. Some alleviation of this may be obtained by retuning the set. The problem may be due to over- loading at the r.f. input in which case an attenuator in the aerial downlead may effect a cure. M Figure 3. Details of the pickup coil. These dimensions should be adhered to for best raeiiltc 238 — elektor february 1975 three-eyed bandit three-eyed From time immemorial man has played games of chance. In primitive societies the men often sat around throwing dice or playing other games while the women did Lhe work. Nowadays, alas, this situation no longer exists, but modern technology has considerably widened the scope of games of chance so that a whole range of 'gaming machines' can be seen today. The most common type of mechanical or electromechanical gaming machine is the ‘one-armed bandit’ or ‘fruit ma- chine’. In such machines three cylinders bearing numerals or symbols are set into rotation simultaneously by pulling a handle or pressing a button. The cylin- ders ultimately stop or can sometimes be stopped by the player and the combi- nation of symbols appearing in a window when the cylinders have stopped deter- mines whether a win has occurred and the magnitude of that win. The less probable combinations are, of course, awarded the higher prizes. The ‘Three-eyed bandit’ described here works on the same principles but is completely electronic. Instead of mech- anical drums there is a display of three columns of four lamps. When a start button is pressed the three columns of lamps flash until individually stopped by three stop buttons. A win is indicated when a row of three lamps is lit and the magnitude of the win depends on which row is lit. Referring to figures 1 and 2 the system Table 1 COUNT D c B A Li L2 *-3 l 4 LAMP LIT 0 0 0 0 0 0 0 0 1 1 0 0 0 1 0 0 1 0 EvlHHs 2 0 0 1 0 0 1 0 0 - Sl-iSfei 3 0 0 1 1 1 0 0 0 Li 4 0 1 0 0 0 0 0 1 l 4 5 0 1 0 1 0 0 1 0 l 3 6 0 1 1 0 0 1 0 0 L2 7 0 1 1 1 0 0 0 1 L 4 8 1 0 0 0 0 0 0 1 L4 9 1 0 0 1 0 0 1 0 l 3 Table 1. The truth table for the lamp decoding. The Boolean functions for the four tamps in a column are as follows: (referred to the BCD out- puts of the 7490) Li = A • B • C L 3 = A • B 1-2 = A • B L 4 = A- B- C + S*5 Figure 1. Suggested front panel layout for the Three-eyed Bandit. Figure 2. Block diagram of the Three-eyed Bandit. Figure 3. The complete circuit of the Three-eyed Bandit. start r-n ®Px ® Pv ® Pz X y 2 red Llx ® Llv ® Liz ® 1 st prize yellow L2x ® L2y ® L2z 2nd blue L3x ® L3y ® L3z ® 3rd green L4x ® L4y ® L4z ® 4th 1640 1 operates as follows. Each column of lamps is driven by an oscillator, a decade counter and a decoder. The oscillators are independent and of different fre- quencies. The three oscillators are started simultaneously by the start button and are stopped individually by three stop buttons. The rate of flashing of the lamps is sufficiently high so that no cheating can occur when pressing the stop buttons. Pressing the start button also resets the counters. The weighting of the various combina- tions is achieved as follows. Referring to the column marked X in figure 1 the de- coding is arranged such that L4X lights 4 times in a cycle of ten pulses from the oscillator, L3X lights 3 times, L2X lights twice and L1X lights once. The decoding is mutually exclusive, that is only one three-eved bandit big ben 95 elektor february 1975 — 239 combination and award prizes or points such that the lower the probability the larger the prize. The operation of the circuit is as follows. Only column 1 will be described as the others are identical. In figure 3 N 3 and N 4 form an astable multivibrator. Nj and N 2 form a set-reset flip-flop. When the flip-flop is reset by P x the output of Nx is low. Pin 10 and pin 12 of N 3 and N 4 re- spectively are held low, the outputs are thus high so the astable will not start. When the start button is pressed the flip- flop is set and the astable starts, thus driving the 7490 decade counter until the stop button P x is pressed. The output of the counter is decoded in accordance with the truth table of table 1. Note that a ‘ 1 ’ under a lamp number indicates that the lamp is lit. Note also that the lamp driver transistors T 2 — T 4 are NPN, whereas Tj is PNP. Therefore when a ‘1’ appears at the output of N I3 Lj will go out, whereas when a ‘1’ appears at the output of N 12 for instance, L 2 will light. Tj thus inverts the output of N 13 as far as lighting the lamp is concerned. Warning: This Three-eyed bandit is in- tended for private amusement only. There are very strict laws in the U.K. governing the use of gaming machines for money and other prizes in clubs, public-houses &c. - Ed. M lamp is lit at any time. It is therefore obvious that when the stop button is pressed it is most probable that L4X will be alight and least probable that LI X will be lit. The respective probabilities are: P L4 = 4/10 PL3 = 3/10 P L 2 = 2/10 P L l = 1/10 This is true for all three columns of lamps. The probability of all three lamps in a row being lit simultaneously is given by the product of the individual prob- abilities thus: P(row 1) = PL1X ’ p LlY ' P L1Z = 1/10 • 1/10 ■ 1/10 = 1/1000 Similarly P(row 2) = 2/10 • 2/10 • 2/10= 1/125 P(row 3) = 3/10 • 3/10 • 3/10 = 1/37 P(row 4) = 4/10 • 4/10 • 4/10 = 1/16 Since the probabilities of all the lamps in row 1 being lit is the smallest row 1 obtains the highest prize. Of course a win need not be awarded just on a complete row. Wins could be awarded for part rows as in a mechanical one-armed bandit where a prize might be awarded for two ‘oranges’ in a row. All that is needed is to calculate the probabilities of a particular big ben 95 Nowadays both mechanical and electronic doorbells playing com- plete melodies are commercially available. The Big Ben plays a striking and well-known melody. Figure 1 shows the circuit diagram of this Big Ben. When bell button Dr t is depressed, RS flip-flop N 6 , N 7 is set. The pulse on the output of N 7 changing from ‘1’ to ‘0’ is passed on via C 7 to a second RS flip-flop N 3 , N 4 . Consequently a logi- cal ‘1’ appears at the serial input of IC t . This ‘1’ is shifted onwards at the first clock pulse and arrives at the A output of IC x . The result is that RS flip-flop N 3 , N 4 is reset via N s . Now the serial input is ‘0’ again. The ‘1’ fed into the shift register now moves forward through shift registers ICx ...IC 3 at the frequency of the clock pulse. After 12 clock pulses the D output of IC 3 will become ‘1’. Via N g this ‘1’ resets RS flip-flop N 6 , N 7 , so that the circuit returns to its steady state. The clock pulses are obtained from an astable multivibrator comprising Nx and N 2 which oscillates continuously. The frequency (tempo of the Big Ben mel- ody) can be adjusted to personal taste by means of C s and C 6 . The output voltages of shift registers IC ( ... 1C 3 are supplied to a voltage- controlled oscillator T], T 2 via the trimming potentiometers P, ... Px 2 and D, ... D, 2 . If so desired, P s , P l0 , D 5 and Dx 0 may be omitted to obtain the required rests in the Big Ben melody. Potentiometers P, ... Px 2 govern the frequencies of the voltage-controlled oscillator Tx , T 2 . To prevent this oscillator from oscillating if none of the shift-register outputs is ‘1’ (in its steady state), capacitors C 3 and C 4 are included in the circuit. The oscillation signal of the voltage- controlled oscillator is applied to a loud- speaker via T 3 . As the latter only switches, it does not dissipate much power, so that no additional cooling is required. Figures 2 and 3 show the lay- out and arrangement of the components on the print. The potentiometers are conveniently arranged, so that final trimming to obtain the correct melody is an easy job. M 240 — elektor february 1975 big ben 95 470 --- 1000*2 l— O 0-* S' T Figure 3. The component layout. Figure 2. The p.c. board for figure 1. Figure 1. The circuit diagram of Big Ben. Components list: Resistors: P 1 - p 12 = 10k R-j ,R 2 - 1 k2 R3,R4 = 5k6 R 5 ,R 6 « 470 R 7 = Ik Rg = 47 £2 Semiconductors: IC-| ,IC 2 ,IC 3 = 7495 IC4JC5 = 7400 T 1 ,T 2 = TUN T 3 = BC 140 Dt ... Di 2 = DUS Capacitors: C-| ,C 2 - 1 50n C 3 ,C 4 = 220n C 5 ,C 6 - 470 ... 1000 /i/6,3 V C7 - 56n ciocki IC2 = 7495 PI P12=10k cos/mos digital ics COS/MOS digital ICs ! COS/MOS is a development of bipolar IC technology and an offspring of the MOS (Metal Oxide Semiconductor). It started with the MOSFET being developed from the universally known junction FET (Field Effect Transistor). The former distinguish themselves from the latter by their isolated gate. The result of this gate isolation is a particu- larly high gate resistance. A drawback is that a static charge can build up on such a gate when the transistor is not connected in a circuit. This charge usually causes the immediate destruc- tion of a MOSFET because the extre- I mely thin isolating layer breaks down. So the handling of MOSFETs calls for special precautions. This also applies to COS/MOS ICs in which MOSFETs are integrated. The integration is such that P+ and N- channel transistors are used alter- nately. Furthermore the switching circuits are integrated symmetrically. The latter two characteristics form the basis for the term COS (Complemen- tary Symmetry). Thus COS/MOS can be briefly described as complementary sym- metrical MOSFET integration. A simple example of a COS/MOS IC construction is given in figure A. Here the dark- shaded area represents the n- (polarized) substrate. The diagonally-hatched area is the metal oxide film on which the electrical contacts are made. These contacts are drawn in deep black. Below the isolating layer at the electrical contact interruptions are the p- and n- layers. The layers are so integrated that the result is a complementary MOSFET pair as shown in figure B. Corresponding to the labelling of figure A, we have the following labelling in figure B: ‘S’ for sources, ‘G’ for gates and ‘D’ for common drain. As can be seen from figure A the integration of an N- channel MOSFET is of a simpler construction than a P- channel. The latter requires an extra p- layer separating the substrate from the two n- layers which lie between the drain and G2 (= gate 2) and the junction between G2 and S2 (= source 2), respect- ively. Of course, the integration of even the simplest COS/MOS 1C is slightly more complex than figure B suggests. Even a common 2-input NAND gate consists of no less than four integrated MOSFETs. Like MOSFETs, every COS/MOS IC must be handled with due care because the inputs (gates) are isolated with respect to the rest of the integrated elektor february 1975 - 241 circuit. Normally the input impedance of a gate is 10 12 L2. As a result a static charge can easily build up if such an IC is kept in a plastic box, for instance. The human body too, is often statically charged. Touching the inputs with a finger can be sufficient to destroy the COS/MOS IC. Therefore the ICs are packed in a kind of expanded plastic containing a highly conductive sub- stance. The connecting pins of the IC are pressed into the expanded plastic. To give the inputs some measure of protection, manufacturers often provide COS/MOS 1C inputs with an inbuilt protection circuit. These circuits are not shown in the circuit diagrams of the ICs. Figure C is an example of an input circuit of a COS/MOS inverter. As can be seen in this figure, the circuit consists of a P- and an N- channel MOSFET. In reality the input circuit is as shown in figure D. Here we see that each gate input protection circuit comprises one resistor and three diodes. The diodes D 4 to D 8 are usually formed in the diffusion process. The gate input protection, however, is added as an extra (a resistor of about 500 12 plus three diodes). In figure D the diode D 3 has a break- down voltage of about 25 V. The breakdown voltage of the diodes Di and D 2 is about 50 V. M 242 — elektor february 1975 mos tap I MOS Solid state circuits are increasingly intruding into fields that were once the domain of electromechanical components, but only recently has it become possible economically to replace the simple electromechanical switch by an electronic system with no moving parts - the Touch Activated Programmer (TAP). In last month's issue a TAP was described which utilised TTL IC's. This month we publish a new design based on COSMOS logic packages. In line with the Elektor policy of con- tinuous development and utilisation of new technologies a TAF has been de- veloped which uses COSMOS IC’s. As will be explained later in the text this offers greater circuit simplicity than the TTL TAP but, since COSMOS prices are higher, this circuit is more expensive than the TTL TAP. Readers thus have two designs from which to choose; a TTL TAP using cheap, readily obtainable components, or a MOS TAP using ‘state- of-the-art’ devices at slightly higher cost. The main advantages offered by the MOS TAP are as follows: — micropower quiescent operation. — excellent noise immunity (typically 45% of the supply voltage). — wide supply voltage tolerance (3— 15 V). — high input impedance (typically 1 0 12 S2) therefore, unlike the TTL TAP, no input buffers are needed. The MOS TAP is based on an RCA COSMOS 1C, the CD401 1 AE, which is a quadruple two-input NAND-gate. The circuit of one of the gates is given in figure 1 . It consists of two complemen- tary pairs each comprising a P-channel FET and an N-channel FET. When inputs A and B are both high (+Vb) the P-channel FET’s are cut off. The two N-channel FET’s are turned on and the output is in the low or ‘0’ state, which is a resistance of 400-800 between point C Figure 1. The circuit of one of the NAND- gates in a CD4011AE. Note the use of com- plementary pairs of P- and N-channel MOSFET's. Figure 2. The pinout of the CD4011AE DIL package. The configuration is different from the 7400 used in the TTL TAP. Figure 3. The 'push-button' is the simplest application of a COSMOS NAND-gate as a touch-switch. If points A and B are bridged by a finger the output of the gate will become ' 1 '. If the contacts are released the output becomes ' 0 ' again. T^ is an emitter-follower to increase the output current capability. Figure 4. The basic element of the multi- position switch is a set-reset flip-flop. It is shown here with two sets of touch contacts, but one of these is replaced by the reset bus RB in the final circuit. Figure 5. The basic configuration of the reset circuit. The monostable N 3 /N 4 produces a re- set pulse when input A is touched. This goes out along the RB bus to reset the other switch positions, which, for simplicity are not shown. Figure 6 . Detail of the output circuitry for one switch position. The functions of the three outputs are detailed in the text. eft eft TS ti it x DC T elektor february 1975 — 243 6 Qa La Ua i ~i supply common. When one or both inputs are low the corresponding P- ihannel FET is turned on and the N- channel FET is turned off. +Vb therefore ippears at output C via the on resistance of the P-channel FET. The pinout of the CD401 1 AE is given in figure 2. Note that it is not the same as the pinout of the 7400 which was used n the TTL TAP. The IC is also available ® a (more expensive) ceramic package is the CD4011AD. This has the same pinout as the plastic-packaged CD4011AE. For those who are unfam- iliar with MOS devices it is worth noting that since the devices are of insulated- gaie construction they should be handled with care as static charges can easily testroy the device. In particular it is recommended that an IC socket be used, neither should the device be plugged into iter removed from the circuit with power applied. The NAND-qate as a 'push-button' The basic principle of the MOS TAP is -lustrated in figure 3. In the quiescent >--:e the inputs of the gate (which for this example are tied together) are held at +Vfc by R, . The output is therefore low If points A and B are bridged by a ringer the input will be held low by the sk.n resistance, which is a maximum of about 2ML2 for dry skin and considerably .-ess for moist skin. The output of the gate will therefore become high. Since COSMOS can supply only 500juA or so output current an output buffer may be required for some applications. The emitter follower T[ provides this. If the output is required to sink current in the -O' state R 4 must be included. R 2 is an input protection resistor for the 1C and C helps improve the transient noise mmunity. This simple circuit is, of course, useless - -etching operation is required so, like the TTL TAP, the MOS TAP is based :t set-reset flip-flops. One flip-flop is employed for each switch position and •he circuit of one such flip-flop is given m figure 4. It operates in the following manner. In the quiescent state the inputs (pin 1 and pin 6) are held high by R, and R 4 . Suppose pin 4 is initially high, then pin 2 is also high. In accordance with the NAND-function pin 3 is low, which means that pin 5 is also low. Pin 4 is therefore high which was our original premise. This is one of the two stable states of the flip-flop. Suppose now that input A is touched. This means that pin 1 is held low. Pin 3 therefore becomes high, and since 6 and 5 are high 4 becomes low. This holds pin 2 low so that even when the touch contact is released the circuit remains in this state. If input B is now touched the circuit reverts to its original state. We thus have a two-position switch. Extension to multi-position switch There are various ways of extending the system. One way would be to use NAND- gates with several inputs to make an ‘n-stable’ flip-flop. One NAND-gate would be required for each switch posi- tion. In practice this would be very cumbersome since an n-position switch would require NAND-gates with n+1 inputs. It would also be impossible to further extend the system once it had been built, and of course a different design of printed circuit board would be required for each different number of positions required. The MOS TAP described in this article uses the same system as that described for the TTL TAP. The switch may be extended to any number of positions using the set-reset flip-flop previously described and the latching operates by using a common reset monostable so that when any contact is touched a reset pulse is produced which cancels all the other switch positions. The principle of operation of the reset circuit is illustrated in figure 5, which shows one switch position plus the reset monostable. When input A is touched the monostable consisting of N 3 and N 4 is triggered and produces a reset pulse of about 50 mSec. which goes out along the reset bus RB to reset any positions that are set. CB is the common reset in- put bus and the switch inputs are con- nected to it via diodes (D[) to isolate them from one another. The reset input connected to R 2 directly is the total reset input which may be used to reset all the switch positions if desired. The output circuitry Before describing the circuitry of the complete TAP it is necessary to clarify some points concerning the output cir- cuits. As can be seen from figure 6 only the Q output of the flip-flop is used. T] is a buffer emitter follower, as described earlier. When the flip-flop is in the reset state T! is cut off: when the flip-flop is set, however, a voltage of +Vb~0.7 V. appears at the Q a output (0.7 V. is the base emitter voltage drop of Tj ). The current which Tj can supply is limited by R 6 and depending on the gain of T! can be between 100 and 200 mA; the Q a output is thus short-circuit proof. The optimum value of R 6 is given by: R 6 = 2 X 10 3 X Vb where R 6 is in ohms and Vb in volts. The total current which can be supplied by outputs Q a and L a together is approxi- mately: ‘Qa + ! La = 5 x 10-* X h FETi The current is in milliamps. hpgj is the common-collector current gain of Tj . Output L a is intended to drive a LED (or lamp) to indicate when a particular switch position is energised. For a typical LED with a voltage drop of about 1.5 V. at 40 mA. R 7 is given by: Since the base-emitter voltage of T| will vary with temperature output U a is provided for applications requiring a stable output voltage (the output voltage of U a is equal to Vb so if the supply is stable output U a will be). Of course this output can only supply about 500 /uA. The complete MOS TAP The circuit of the complete five-position MOS TAP is given in figure 7. As can be seen from the circuit each input is con- nected to the common reset input bus CB via a diode (D t — D s ). These isolate the inputs from one another. N 3 and N 4 are the reset monostable. The system may be extended simply by adding extra boards. In this case, since only one reset monostable is required for the entire system, N 3 and N 4 on the additional boards may be converted to extra switch positions by adding the components shown in the dot-dash lines, i.e. D x , T x , R a , Rb, Rc- The components shown in dotted lines (i.e. Rj, Ci and the links across D x and between pin 1 1 of the 1C and pins 1 and 2 of the board) are omitted. Note the new link between Figure 7. The circuit of the complete MOS TAP. This has five switch positions, but if the system is extended the monostable on ad- ditional boards can be converted to a flip-flop, thus providing six extra positions per board. Figure 8. The printed circuit pattern for the MOS TAP. Figure 9. The component layout of the basic MOS TAP. elektor february 1975 — 245 Parts list for figures 7, 9 and 10. Resistors: Rl = 2M7 R 2 ,R4.Rg, R 14.Fil9. R 24 = 1M R 3. R 5- R 10' R 15< R 20- R 25 = 10M R 6 * R 8 ' R 1 1 - r 13- r 16' R 18- r 21 • r 23> r 26 > R 28 = 27k* r 7- r 1 2 > r 1 7' R 22- R 27 = 220ft* Capacitors: Cj = 470p C 2 « 47 n C a -C e = lOOp* Semiconductors: ICi ,IC 2 ,IC 3 = CD401 1 AE or CD401 1 AD (RCA) Tt -T 5 =BC109b or BC109c D-| — D 5 = BA127* or equivalent For the extension board C-) and Ri are not required. The following additional parts are needed. R a ,R c = 27k* R b = 220ft* D x = BA127 T x = BC109b or BC109c * See text. Figure 10. The extension board component layout. Note the differences between this board and figure 9. Photo 1. The completed five-position MOS TAP. pins 2 and 3 of the board, and the link which replaces . The printed circuit board The board for the basic MOS TAP is given in figure 8 and the associated com- ponent layout in figure 9. The board layout for extending the system is shown in figure 10. The component differences to the left of ICi can be clearly seen. It can be seen that the supply tracks +V b , 0 V. and the reset lines RB and CB are available on both edges of the board, so that extending the system is simply a matter of linking across. The capacitors C a — C e and C x are included to improve the transient noise immunity, but they may not be required in every case. If it is desired that the switch should set in a particular position on switching on the supply than all these capacitors should be omitted, except the one connected to that switch position. Note that diodes D! — D s and D x must have very low reverse leakage, less than 200 nA., so DUS cannot be used in this circuit. If these diodes are omitted then the CB rail has no effect. In that case any number of switch positions may be on simultaneously and they can only be reset by touching the reset input con- nected directly to the reset monostable. Power Supplies The MOS TAP will operate from any supply between 3 and 15 V. The current consumption is very low, less than 10 nA. at 1 5 V. but of course any output current the circuit must supply is added to this. Precautions As mentioned earlier COSMOS IC’s must be treated with extreme care, and in particular the use of an IC socket is recommended. If the device is soldered directly into the circuit use an earthed soldering iron. Touch contacts The design of the touch contacts is up to the user, but should be such that they cannot be accidentally bridged by dirt, moisture etc. When a supply of less than 10 V. is used it is generally possible to employ single-point touch contacts (no earth return) as the circuit will operate from hum picked up by the body capacitance rather than from the skin resistance. Screened leads should be used in both cases if the input lead length exceeds 5 cm. and the screening should be connected to supply common (0 V.) at one end only. N i 246 — elektor february 1975 modulation systems modulation The recent energy crises have | I 1^^ underlined the need to forge ahead J with the development of new communication systems, not only to alleviate the ever-increasing wave- band congestion, but also to make more economical use of transmitter power. It is hoped that this article will give an insight into the various modula- tion systems now in use and will also explain the design philosophy of a new transceiver developed by Elektor. ; fa XEC Communication systems The purpose of a communication system is to convey information from one loca- tion to another (distant) location. The block diagram of a communication sys- tem is given in figure 1. It comprises three parts: — an encoder to convert the information into a form suitable for transmission via the medium. — the medium. — a decoder to convert the information back into its original form. One of the oldest communication sys- tems utilises the human voice. Informa- tion from the brain is encoded into mechanical vibrations by the vocal sys- tem, transmitted via the air and recon- verted by the aural system of the listener into information in the brain. This system, although still widely used, has its drawbacks. Notably that the range is limited by the power of the lungs. The system is also subject to inter- ference from nagging wives, mothers- in-law etc. and prone to breakdowns due to laryngitis and other complaints. As another example of a communication system consider the postal system. In- formation from the brain is encoded in the form of writing, transmitted via the postal system (the medium), and de- coded by the optical system of the recipient. These two examples both require direct human intervention in the transmission and reception of the information, but this is not always necessary. Two com- puters, connected by a data link could carry on a meaningful dialogue without human interference or an unmanned meteorological station might transmit data to a remote terminal. Communica- tion systems may therefore be divided into at least two categories: — systems in which information per- ceptible to human sense organs is transmitted and in which the ultimate receiver is a human sense organ which decodes the information in conjunc- tion with the brain. Radio broad- casting and television fall into this category. — systems in which human senses play no part in the decoding process. The difference between the two stems from the fact that human senses can operate very selectively so that the de- sired information can be extracted in the presence of large amounts of un- wanted information (noise & c.). This faculty may be further improved by training, so that a radio operator can frequently distinguish signals that would be unintelligible to the layman. It is thus possible to subdivide the first category into two sub-groups: — systems in which the impairment of the transmitted information must be as small as possible, for instance high-fidelity f.m. broadcasting and television. b ra. m fa m ' TO: a - systems whose functional capability is little impaired by distortion of the information content or even by omis- sion of a large part of it. I Systems such as the telephone, which in I general convey only speech, fall into this I category. Speech is still intelligible even I after removal of a large amount of the I information by restricting the bandwidth I or by other means. nil, which occurs on the ‘troughs’ of the modulating waveform. It therefore fol- lows that if the modulation is linear the maximum amplitude is twice the ampli- tude of the unmodulated carrier on the peaks of the modulating waveform. A DSB signal is shown in photo 1. The mathematical expression for this form of modulation with a sinusoidal modulating signal is as follows: v=[l+m-cos(co AF t)]-v 0 cos(co RF t) (l) where m is the modulation index v 0 cos(w RF t) is the carrier, of which v 0 is the peak unmodu- lated value cos(co AF t) is the modulating signal. are not single frequencies but a spectrum of frequencies occupying a bandwidth equal to ± the highest modulating fre- quency on each side of the earner. It can be seen from the equation that even with a modulation index of 1, half the energy radiated is at the carrier fre- quency and contains no information. In fact commercial broadcast transmitters operate with a mean modulation depth considerably less than 1 00%. It is there- fore apparent that transmitter power amounting to many gigawatts is being radiated uselessly into space by trans- mitters around the world. Apart from the waste of energy other undesirable phenomena occur, such as cross-modula- tion in the ionosphere (the Luxembourg effect). Furthermore, the system is in- efficient in its use of bandwidth since it uses two sidebands each containing the complete LF information. One of these is clearly redundant. It seems legitimate to ask why, in the face of all these objections, DSB is the most common system in use at the pres- ent time. There are two reasons. Firstly it has the stamp of antiquity. DSB is the oldest modulation system in use and consequently much capital is invested in transmission and receiving equipment. Secondly, it is the simplest system to implement, whereas more economical systems (in terms of power and band- width) are considerably less economical in terms of equipment cost, though viewed in the long term not unduly so. The circuit of a simple modulator is shown in figure 4. The supply potential of the transistor oscillator, and hence its output, is varied by the output of the modulation amplifier. Demodulation, or detection, of a DSB signal is simply accomplished by means of a diode and a low-pass RC filter. The diode rectifies the modulated waveform so that only the negative half-cycles appear at its cathode. This output con- tains one-half of the original envelope, that is the original modulating signal. The carrier is simply removed by the low-pass filter and only the original modulation appears at the output super- imposed on a d.c. potential correspond- ing to the amplitude of the unmodulated carrier. To increase the useful radiated power of DSB transmissions dynamic range compression is often employed. This means that the range between the loudest and softest sounds of the modu- lating signal is reduced, or to put it another way, pianissimo is boosted and fortissimo reduced. This means that the variation in the modulation depth is re- duced. A simple compressor is shown in figure 6. The signal at the collector of the transis- tor is rectified by D,/D 2 and the poten- tial on the capacitor at point A is applied to the base of the transistor to provide bias. If the signal through the transistor is increased the potential at point A de- creases, reducing the base bias of the transistor so that the working point is shifted to a point where the gain is less. Choice of suitable time constants in the The concept of Modulation When electromagnetic radiation serves as a medium for the transmission (i.e. as a carrier) it is necessary to impress the information onto the carrier by changing one or more of its parameters (Le. to modulate it). The decoder at the receiving and reconverts these changes into in- formation. If the discussion is confined to analogue modulation there are two important types: - amplitude modulation (AM) - frequency modulation (FM) In amplitude modulation the variable parameter representing the information is the amplitude of the carrier, whereas for frequency modulation the variable is the frequency of the carrier. The more important forms of amplitude modulation are as follows: - double sideband with carrier (DSB) - double sideband suppressed carrier (DSSC) - single sideband suppressed carrier (SSB) - carrier position modulation (CPM) The modulation index can have values between zero (no modulation) and unity (maximum modulation). The depth of modulation is frequently expressed as a percentage in which case 1 00% corre- sponds to a modulation index of 1. In commercial broadcast transmitters the depth of modulation is around 30%, which occurs when the AF signal reaches its maximum value, that is cos(co AF t) = 1 . The mean value must therefore necessari- ly be lower. Multiplying out equation (1) gives: v=v 0 ‘cos(ajR F t) + • [cosIwrjH-goaf)* + cos(w RF -co AF )t] From this equation it can be seen that the low-frequency information appears in two sidebands, placed symmetrically above and below the carrier frequency. Figure 3 shows the frequency spectrum of a DSB signal. Of course with a com- plex modulating waveform the sidebands Figure 1. Block diagram of a communication system comprising an encoder, transmission medium and encoder. 2. When electromagnetic radiation as the transmission medium encoding dace in the transmitter and decoding in Figure 3. The frequency spectrum of a DSB signal with single frequency sinusoidal modu- lation. Figure 4. DSB signals may be generated very simply as this diagram shows. Figure 5. Simple DSB demodulator (envelope detector) using a diode as the non-linear element. ► cl * 1 1 L r a — 6018 ' & 248 — elektor february 1975 modulation systems rectifier circuitry means that the dis- tortion unavoidable with compressors is reduced to a minimum. Where a communication system is to be used for speech only, intelligibility is more important than fidelity and large amounts of distortion may be tolerated. Modulation depth may then be con- trolled in a much more effective way by ‘clipping’. In figure 7 the maximum out- put of the AF amplifier is limited to the forward voltage of the diodes, so any peaks in excess of this are clipped. After clipping the signal contains a lot of harmonics, so a low-pass filter is included to limit the bandwidth and thus make more economical use of waveband space. Double Sideband Modulation with Suppressed Carrier (DSSC) As the name suggests, with this type of modulation only the information-carry- ing sidebands are radiated. The waveform produced with a sinusoidal modulating signal using this type of modulation is shown in photograph 2. Suppressing the carrier obviously saves a great deal of transmitter power, but it is apparent from the photograph that the envelope of the resulting waveform is not the original modulating signal, which makes detection more difficult. Generation of DSSC signals is accomplished fairly easily by a number of circuit arrangements, Figure 6. A dynamic compressor is simply an amplifier whose gain decreases as the input signal amplitude increases. Figure 7. Block diagram of an AF clipper using a pair of diodes. Figure 8. A symmetrical balanced modulator for the production of DSSC signals using an 1C. The components in the 1C are shown in the shaded portion of the diagram. Figure 9. A simple product detector which may be used where good rejection of the input signals is not essential, or where this is done elsewhere in the circuit. Figure 10. A 'universal' demodulator which will demodulate all existing narrow-band mo- dulated signals both AM and FM. Figure 11. Generation of an SSB signal by the filter method. the most effective being the symmetri- cal balanced modulator, of which a full range is available in 1C form. The circuit of a typical balanced modula- tor using such an IC is given in figure 8. Ti/T 2 and T 3 /T 4 form two differential pairs. The carrier is fed in through the input transformer, but in the absence of a modulating input the outputs of the two differential pairs cancel and no carrier appears at the output. T 5 and T 6 form a differential pair into which the modulating signal is fed. This causes the pairs Ti/T 2 and T 3 /T 4 to deviate from the balanced condition and a signal appears at the output which is pro- portional to the product of the LF and RF signals. i e. V out = V, • V 2 Demodulation of DSSC signals is ac- complished by means of a product detector whose output is the product of the two input voltages. So that for Vj = cosCcoaf 1 ) ‘ coslcjRpt) (the DSSC signal) and V 2 = COS(CORpt+V>) (the regenerated carrier) the output becomes V out = cos(waf 0 * cos(ojRpt) ■ COS(C0Rpt+v5) = icOs(lOAFt) ■ [cos>p + COs(2WRpt+AFt) ' cos <{> (3) The restored carrier required for de- modulation can be derived from the sidebands but the practical difficulties are considerable. For this reason a small fraction of the original carrier is radiated as a pilot frequency to facilitate re- generation of the carrier at the receiving end. In the receiver this so-called residual or vestigial carrier has its level raised to the value required for demodulation. A phase-locked loop system is used because of the stringent phase criteria which have to be met. For instance, looking at equation (3) it is apparent that if the re- stored carrier is shifted in phase by 90° from the original carrier the LF output will be zero. Although it is easy to achieve the correct phase with several systems, a PLL system is one of the few which will maintain a phase relationship with time and temperature changes. For the product detector the type of IC used for a balanced modulator may also i be used. As the high-frequency com- I ponent is quite easy to suppress at the output less complex circuits are generally j used for demodulation. The more simple j circuits are, however, prone to fading and RF interference. In the Elektor laboratories a number of experiments were carried out to compare the performance of various product de- tector circuits and it was found that IC’s with circuitry similar to figure 9 gave the best results. Figure 10 is the block diagram of a de- modulator for DSSC signals. When a signal is tuned in the phase-locked loop regenerates the vestigial carrier. Since most practical PLL’s operate with a 90° phase shift the signal is shifted by 90° before being fed to the product detector. The output of the product detector is fed to a low-pass filter which removes the high-frequency components. This system will also demodulate normal DSB signals and although it may seem a little over-engineered compared with a diode detector it does offer significant ad- vantages, particularly in the presence of interference. Furthermore, it will become apparent that this system will de- modulate SSB and CPM signals plus fre- quency and phase modulation. In fact it is a universal demodulator for all practi- cal forms of analogue modulation. Single-sideband Modulation with Suppressed Carrier (SSB) In DSSC modulation the carrier (which contains no information) is suppressed to save transmitter power, but this makes no economies in the bandwidth required to transmit the information as compared to DSB. The transmitted signal still has two sidebands above and below the carrier frequency and since each sideband contains all the LF information one of them may be discarded with a conse- quent halving of the required bandwidth. This is what happens with SSB, hence the name. For any given bandwidth twice as many SSB transmissions may be accommodated as compared with double-sideband transmissions. There are various ways of generating an SSB signal. The simplest way is to start with a DSSC signal and to suppress one of the sidebands by filtering so that only one sideband appears at the filter output. This method offers the choice of ra- diating either the upper sideband (USB), or the lower sideband (DSB) depending on the choice of filter parameters. The filter method of generating an SSB signal is shown in figure 1 1 . Since the two sidebands are separated only by twice the lowest frequency of the mo- dulating signal the filter must have a sharp cutoff if adequate rejection of the unwanted sideband is to be achieved. Since filter slopes are quoted in terms of dB/octave (an octave above or below a frequency is twice and half that fre- quency respectively) it follows that the lower the carrier frequency the further apart are the sidebands in terms of octaves, and the easier it is to filter out the unwanted sideband. For this reason the signal is often modulated onto a carrier frequency much less than the transmitter frequency and after filtering out the unwanted sideband the fre- quency is raised by frequency conversion to the frequency to be transmitted. Carrier suppression is also easier at low frequencies. Currently available ceramic filters can give up to 50dB rejection of the unwanted sideband. A second method of generating SSB signals is shown in figure 12, but is less common. In this arrangement the LF signal is split into two components with equal amplitude but with a 90° phase shift with respect to one another and the carrier is dealt with in a similar fashion. The LF and RF signals are then fed to two balanced modulators. The two DSSC signals so produced are dis- placed in phase so that if a sideband of one signal is in phase with the corre- sponding sideband of the other signal then the other two sidebands will be 180° out of phase. Adding the two DSSC signals will therefore cancel one sideband, and subtracting them will cancel the other, so the desired sideband may easily be selected. Since accurate wide-band phase-shifters are often diffi- cult to realise in practice a third method of producing an SSB signal is shown in figure 13. This is a two-stage modulation procedure. The LF signal is modulated onto two sub-carriers displaced in phase by 90°. The upper sidebands are rejected by the filters and the two signals are then processed as were the LF signals in figure 1 2. Speech Processing The envelope of an SSB signal bears no resemblance to the original modulating waveform and attempts to raise the average level of the transmitted signal by I low-frequency processing such as com- pressors or speech clippers are doomed to failure. The most effective way of raising the average level of an SSB trans- mission is by limiting of the RF signal itself. This causes harmonics, widening the frequency spectrum so that the limit- er must be followed by a filter to remove - them, if the SSB signal is produced by the filter method then a similar filter may be I used for the removal of the harmonics I after clipping. Figure 14 shows the block diagram of a typical RF clipper. These I devices are available commercially in various forms under the name ‘Speech _ ~ Processor’. Figure 12. The phase method of producing an SSB signal avoids the use of steep-slope filters. Figure 13. A 'double modulation' system for the generation of SSB signals. Modulation of the AF signals onto a sub-carrier avoids the use of wide-band phase shifters. Figure 14. An RF limiter 'speech-processor' which increases the average radiated power. Figure 15. Demodulation of an SSB signal with a beat-frequency oscillator (BFO) and product detector. modulation systems maxi display elektor february 1975 - 251 Demodulation An SSB signal may be demodulated with the system of figure 10, but if speech only is to be transmitted the simpler arrangement of figure 1 5 may be used. This system will not work satisfactorily with DSSC signals due to beat frequency problems. For example, if the regener- ated carrier is not synchronised to the incoming carrier but is displaced in fre- quency by say 100 Hz then the sidebands will also be displaced by this amount but in opposite directions. This results in the production of a strong 200 Hz com- ponent which renders the signal quite unintelligible. With an SSB signal the only result would be to displace all the frequencies by 1 00 Hz and speech would probably still be intelligible, although this would be useless for music. Interference with AF Equipment All AM transmissions possess one com- mon characteristic, that there is a corre- lation between the amplitude of the radiated signal and the amplitude of the modulating signal. Feeding such signals through an envelope detector will there- fore result in an AF output, though with DSSC and SSB this will be unintelligible. In principle any non-linear element will function as an envelope detector pro- vided the amplitude of the RF signal is sufficient. For this reason interference with domestic electronic equipment such as television and Hi-fi equipment can be a problem. Any of the semiconductor junctions in such equipment (and even dry joints, dirty plugs and the like!) could demodulate an unwanted RF signal although the most frequent cause of trouble is in the input stages of Hi-fi amplifiers. Radio amateurs are often unjustly blamed for such interference, but their equipment generally complies with the regulations and the fault is in the design of the equipment in which the interference is occurring. Constant Amplitude Systems Transmission systems in which the am- plitude of the carrier remains constant rarely give rise to interference in dom- estic equipment. One possible exception is where an audio amplifier is blocked completely by a strong RF signal, but this occurrence is rare. Such systems are not necessarily more wasteful of trans- mitter power, since when used for speech voice-operated switches may be used so that the transmitter operates only when an LF signal is present. Another advan- tage of constant-amplitude systems is that automatic gain control (AGC) is much easier to include and indeed may sometimes be omitted altogether. The second part of this article will deal with the characteristics of carrier posi- tion modulation (CPM), frequency and phase modulation. (To be continued) M maxi With LEDs now available for a few tens of pence each it is possible to construct a 'giant' seven-segment display from discrete LEDs for the price one would normally pay for a medium sized integrated display. This maxi-display uses 22 LEDs per digit (3 per segment plus decimal point) and incorporates a decoder. As can be seen from the circuit diagram this display incorporates the improved 6 and 9 format described elsewhere in this issue. No current-limiting resistors are included with the LEDs since about 4.5 V of the supply voltage is dropped due to the LEDs forward voltage and their own internal resistance is sufficient to limit the current from the remaining 0.5 V to a safe value. The supply voltage must not exceed 5 V, however, when the current per segment will be about 25mA. A double-sided p.c. board is used for the construction of this display with the decoder components mounted on one side and the LEDs on the other as shown in the accompanying diagrams. r I 252 — elektor february 1975 recip-riaa recip-riaa The performance of bought or self-built preamplifiers for magnetic pickup cartridges is invariably not sufficiently well known. This is mainly due to the work involved in accurately measuring the amplitude response (RIAA or I EC curve), overdrive margin, distortion, signal-to-noise ratio and hum level. When the reproduction quality is not quite what it should be, the blame is by established tradition laid at the door of the disc manufacturer - or if his product is demonstrably above suspicion, at those of the cartridge or loudspeaker makers. The simple (and above all, electronic) preamplifier 'will surely not be misbehaving?' The weighting network described in this article greatly simplifies the above-mentioned measurements. Despite its simplicity, using only five components, it will deliver a measurement signal that is within 0.2dB of the standard RIAA cutting-curve. This should make it just about the smallest professional test instrument ever described . . . During the cutting of gramophone rec- ords, optimum use of the possibilities of groove modulation requires that the lower audio frequencies be attenuated (relative to mid-range) and that the high- er frequencies be emphasized. To enable a flat playback response to be readily obtained, this weighting is done accord- ing to an international (IEC) standard — the former RIAA-curve (figure 2). When the preamplifier amplitude-frequency re- sponse is the inverse of the cutting-curve, the overall response will be correct. Figure 1 shows this playback equalisa- tion curve. Carrying out measurements on the pre- amplifier now involves two specific, normally time-consuming complications. First of all, one cannot straightforwardly check the frequency response. The re- sponse of, for instance, a power ampli- fier should be ‘flat’. This can be quickly checked by applying a constant voltage from a low distortion sine-wave oscil- lator, then observing the more or less stationary pointer of the output voltage meter as the oscillator is tuned through the audio range. By contrast, carrying out such a check on the dynamic pre- amplifier requires a point-by-point com- parison of the meter reading with a voltage-frequency table to see if the figure 1 curve is accurately produced. (See table 1.) This brings us to the second complica- tion. A correct test of the nominal or maximum available output voltage as a function of frequency is only obtained when the input voltage follows the weighting curve used during cutting (figure 2). Allowing for typical levels of disc modulation and cartridge sensitivity, this measurement can be carried out using table 2 — which should cause the circuit to deliver a constant output voltage. The ‘OdB reference’ level used in this table is 5mV at 1 KHz. The input voltages specified to achieve reference output level at other frequencies follow curve 2. The simple and direct solution to both problems should now be obvi- ous: insert a weighting network having the amplitude-frequency response of figure 2 between the constant-voltage oscillator and the preamplifier under test. The input voltage will now vary with frequency according to table 2, whereby the 1 KHz reference level can of course be chosen according to indi- vidual requirements. If the preamplifier is doing its job properly, it will now deliver a constant output voltage — which can easily be checked. The weighting network The RIAA (IEC) characteristic is defined by the time-constants = 75/iS, r 2 = 318/tS, r 3 = 3180fiS. t 2 is opposite in slope to the others. The response of the equaliser-preamplifier (curve l)must be: „ , (1 + P r 2 ) e(p) (1 + pr, ) ■ (1 + pr 3 )’ The cutting system, and therefore also the network described here, has the reciprocal transfer function: H c (p) = (1 + pTi) • (1 + pt 3 ) (1 + pr 2 ) It is not difficult to design a single net- work which will show response break- points corresponding to the three time constants. With the network shown in figure 3, compensation for mutual inter- actions requires the use of three rather different RC-time constants: r, = R, • C, = 82jiS; t 2 = Ri • C 2 = 240jxS; t 3 = R 2 • C 2 = 3000^S. The network actually also has a fourth break-point, which causes the character- istic to flatten off at a frequency well above the audio range. This frequency is near 50 KHz, corresponding roughly to the time-product of R 3 and C | (about 3#iS). The time-constants specified above are chosen to give the best fit of the IEC- curve using standard component values from the E24 (5%) range. The actual component values given in figure 3 fit particularly well. Use of 1% tolerance components will provide an inaccuracy of less than 0.2dB. theoretically, while Table 1. Numerical values for the IEC (RIAA) playback curve. A preamplifier supplied witha constant input level should deliver these out- put levels. Table 2. Numerical values for the IEC (RIAA) cutting curve. A preamplifier supplied with these input levels should deliver a constant output level. Table 3. Comparison of the theoretical IEC/ RIAA cutting curve and the response of a prototype weighting network. The error is less than 0.1 dB from 40 Hz to 16 KHz - although this particular unit was assembled from 5% components! Figure 1. The IEC/RIAA playback equalisation curve. For CD4 and UD4 carrier-channel discs the curve flattens off at 20 KHz, due to an extra time-constant of approx. &[£. Figure 2. The IEC/RIAA weighting curve used during disc-cutting. The 'recip-RIAA' network also produces this curve. Figure 3. Circuit diagram of the network. C 2 can be made up by parallel connection of twice 1.5nF (or 2.2nF plus 820pF). Figure 4. The measurement set-up. The weight- ing network (WN) is inserted between the LF generator (GEN) and the disc-preamplifier under test (DP). An AC millivoltmeter (mV) can now be used to check if the output voltage is the same for all audio frequencies. Figure 5. Determination of drive-limits for disc cutting is a complex matter. The contours given here apply to the stylus tip velocity for the innermost grooves (diam. approx. 140mm) without use of a tracing-distortion compensa- tor. Such devices allow far higher treble levels to be cut and are mandatory for carrier-discs. Figure 6. The inner-groove maxima (figure 5) as they appear after equalisation (at the preamp output). The outer grooves will take +14dB from 50 Hz to 4 KHz. The preamp must have a further overload margin, particularly at higher frequencies, to allow for tracing dis- tortion compensator cuttings. recip-riaa elektor february 1975 — 253 Table 1 Frequency (Hz/KHz) Output level (dB> (mV)* 20 (Hz) 19.3 923 30 18.6 851 40 17.8 776 50 17.0 708 60 16.1 638 80 14.5 531 100 13.1 452 200 8.2 257 300 5.5 188 400 3.8 155 500 2.7 136 600 1.8 123 800 0.8 110 1 (KHz) 0.0 100 (ref) 2 - 2.6 74 3 - 4.7 58 4 - 6.6 47 5 - 8.2 39 6 - 9.6 33 8 -11.9 25 10 -13.7 21 16 -17.7 13 20 -19.6 10.4 * millivolt table based on OdB = lOOmV; change of reference level means that all values have to be changed by the same factor. Table 2 Frequency (Hz/KHz) Input level (dB) (mV)* 20 (Hz) -19.3 0.54 30 -18.6 0.59 40 -17.8 0.64 50 -17.0 0.71 60 -16.1 0.78 80 -14.5 0.94 100 -13.1 1.11 200 - 8.2 1.95 300 - 5.5 2.65 400 - 3.8 3.23 500 - 2.7 3.66 600 - 1.8 4.06 800 - 0.8 4.56 1 (KHz) 0.0 5.00 (ref) 2 2.6 6.7 3 4.7 8.6 4 6.6 10.7 5 8.2 12.9 6 9.6 15.1 8 11.9 19.7 10 13.7 24.2 16 17.7 38.4 20 19.6 47.7 * millivolt table based on OdB = 5mV. Table 3 Frequency (Hz/KHz) 1 EC/ RIAA curve (dB) proto- type (dB) error (dB) 20 (Hz) -19.3 -19.1 + 0.2 30 -18.6 -18.4 + 0.2 40 -17.8 -17.7 + 0.1 50 -17.0 -17.0 0.0 60 -16.1 -16.0 +0.1 80 -14.5 -14.4 + 0.1 100 -13.1 -13.0 + 0.1 200 - 8.2 - 8.2 0.0 300 - 5.5 - 5.5 0.0 400 - 3.8 - 3.8 0.0 500 - 2.7 - 2.6 + 0.1 600 - 1.8 - 1.8 0.0 800 - 0.8 - 0.8 0.0 1 (KHz) 0.0 0.0 0.0 2 2.6 2.6 0.0 3 4.7 4.8 + 0.1 4 6.6 6.7 + 0.1 5 8.2 8.2 0.0 6 9.6 9.6 0.0 8 11.9 11.8 -0.1 10 13.7 13.7 0.0 16 17.7 17.6 -0.1 20 19.6 19.4 -0.2 ■iiiiiiii ■ilium ■■nun ■i:^!lll iiiih.: ■iiiiiiii ■iiiiiiii ■iiiiiiii ■iiiiiiii ■iiiiiiii IMIIIIIII l■III^IIH■IIIIIIH■li l■lllllllMlllllli■■li nillllllM^MilMSIll ^■iiiiiiHMiiiiiiaHaiiii ia£:!!'iiMiiiiiiimiHii 11111111*239111 llllllimikZMIllMHi HIliM IIIIIIII imiiiii m: met) need not be an objection. Measurement procedure Figure 4 shows the measurement set-up. The weighting network is inserted be- tween the LF oscillator output and the preamplifier input. It is good practice - to avoid HF breakthrough, if for no other reason — to arrange that the signal-return is inside the cable screening sheath. This means the use of multiple- core screened cable, with the screen earthed at only one (either!) end. Far too much audio wiring uses the screen as a happy-dumping-ground for returning signal currents, simply asking for ( and frequently getting) HF breakthrough and hum. The reference level of disc modulation - the ‘OdB level’ - corresponds (for CD4 and UD4 carrier discs) to a stylus tip velocity of 22.4mm/sec (peak value) at 1 KHz. Normal Ip’s (and some Ameri- can CD4 discs) have a reference level about 5dB higher. This level generally corresponds to the average level in the loudest passages although the instan- taneous peak programme level can be considerably higher, perhaps +10dB for Table 4 Maximum amplitude error with 1 % components Components and the PC board The PC board is designed to enable DIN connectors to be soldered onto it directly. As mentioned above, optimum accuracy will be guaranteed only when compon- ents of 1% tolerance are used - typicall) metal film resistors and polystyrene capacitors. With almost any available LF oscillator, however, use of carbon film resistors and normal ceramic capaci- tors will not degrade any other aspect of total performance. The simplest ap- proach is therefore to use readily avail- able 5% components, either selecting values with a bridge or else accepting the risk of ± 0.5dB inaccuracies. Fev are better than this anyway Maximum amplitude error with 5% components Signal-to-noise ratio Distortion (with film resistors) below noise Oscillator output voltage for routine testing (OdB = ±3.5mV at 1 KHz) lOOmV Oscillator output voltage for overdrive test (+26dB) preamplifiers The PC board has positions which enabli C 2 to be made up with two component (e.g. 2 X 1 -5nF or 2.2nF + 820pF), sinci the ‘in-between’ E24 values are no always easy to obtain. H how to gyrate — and why elektor february — 255 how to gyrate - The gyration principle was suggested by theoreticians over 25 years ago, although it is rarely seen in practical circuits. It can, however, be used to simulate an inductance of (say) 10,000 H with a Q of 100 in a volume of less than one cubic inch . . . ! In this article the theoretical principles of the gyrator are discussed, and some practical circuits and applications are presented. To be able to understand and use gyra- tors, a certain amount of theoretical background knowledge is necessary. The basic circuit consists of two ampli- fiers (figure 1), with the input of one connected to the output of the other and vice versa. Amplifier A is an inverting and amplifier B is a non-inverting type. The slope of amplifier A is s , = _ g , (A/V) and the slope of amplifier B is s 2 = g 3 (A/V). This means that if amplifier A is driven with an input voltage v, volts, it will de- liver a current of -gj • Vi amps; in other words, it will sink a current of g! • v, (A). Referring now to figure 1 it is clear that the voltages and currents are defined by the formulae: h =gi ‘ v, (amplifier A; the current into this ampli- fier is defined as positive, so that the minus sign disappears); and •i = g2 * v 2 (amplifier B). In these formulae gi and g 2 are so-called gyration-constants. They are very often equal (gi = g 2 = g); sometimes the phrase “gyration resistance” is used, de- fined by In figure 2 the recognised symbol for a gyrator is shown. The next step is to connect an impedance (Z j ) across one set of terminals (dotted in figure 2). In this case the ratio of V! to ii is determined : v, =i, • Z,. From the gyrator formulae it is obvious that the voltage and current at the other set of terminals are defined by: 12=51 * Vi, and v 2 . h g 2 * This means that the impedance “seen” across this second set of terminals is: : ^2 12 Il/g2 _ 1 gl gi ’ g 2 • Zi ( 1 ) What a gyrator does The most important application in prac- Figure 1. Block diagram of the basic gyrator circuit, consisting of a non-inverting and an inverting amplifier. Figure 2. The recognised symbol for a gyrator; the function is to "gyrate" an impedance Zi across one pair of terminals to a different (virtual) impedance (Zj) across the other pair of terminals. tice is the simulation of inductors, for use in LC resonant circuits and the like. If the impedance Z\ in figure 2 is a pure capacitance: z - -sie then the previous formula shows that the virtual impedance across the other set of terminals (Z 2 ) equals: Z, = = JCO; jooC • gl • g2 gl ’ g2 ( 2 ) In words: if a capacitance is connected to one set of terminals, the other pair of terminals behave as if an inductance were connected between them with a value in Henries equal to the capacitance in farads divided by the product of the gyration constants. The gyration con- stants themselves are equal to the slopes of the two amplifiers, which leads to the interesting conclusion that a lower value for the slope leads to a higher value for the simulated inductance! It is apparent that an LC (parallel) resonant circuit can be simulated with the circuit shown in figure 3a. In this circuit the resistors R! and R 2 each represent a parallel connection of the resistive components of the input im- pedance of one amplifier and the output impedance of the other amplifier (and the leakage resistance of the capacitor, which is usually negligible). From the general gyrator conversion formula ( 1 ) it can be shown that this circuit is equivalent to the circuit in 256 — elektor february how to gyrate — and why figure 3b. This, in turn, is equivalent to the circuit in figure 3c provided R] and R 2 in the original circuit are sufficiently large compared to the impedance of Cj and C 2 at the operating frequency. The components in figure 3c are derived from those in figure 3a as follows: gi • g2 From these values the resonant fre- quency (f 0 ) and the quality factor (Q) can be calculated: ^2 “ ^ 2 ; c 2 =c 2 . 27T\/Li • C 2 Q “ 27rf 0 C 2 gl ' g2 Rp + R2 Rp • R 2 . / 6l • g2Ci Rp +R 2 V C 2 In practice one can usually substitute gi = g 2 = g, and Ci = C 2 = C, so that the formulae simplify to: ,4a) If furthermore Ri = R 2 = R, 4a can be further simplified to: Q = jg • R. Summary When an impedance (Z^ is connected across one set of terminals of a gyrator, a virtual impedance (Z 2 ) appears across the other set of terminals: Figure 3. The most important practical applica- tion of a gyrator: simulating an LC parallel resonant circuit (figures 3B and 30 with the aid of a gyrator and two capacitors (figure 3A). Figure 4. In the same way a series resonant circuit (4B) can be simulated with a gyrator and two capacitors (4A). Figure 5. The basic circuit (5A) and a block diagram (5BI of the one tun gyrator. Figure 6. A practical application of the one tun gyrator in a smoothing circuit for the supply rail of a preamplifier. The impedance of the section inside the dotted lines is equiv- alent to an inductance of 250 H! Figure 7. The circuit of the gyrator which is used in the minidrum. If a short pulse is applied to input P a decaying 30 Hz sine wave appears at the output, simulating the sound of a bassdrum. Figure 8. An oscilloscope photo of the bass- drum in operation. The upper trace is the pulse applied to the input and the lower trace is the output. (Scale: horizontal 50ms/div., vertical 200mV/div.) IU ITl riTi*l.il m I kJ% 11 til 1 ill 1 1 LiTITHBH TT 11 T how to gyrate — and why elektor february — 257 z 2 1 gi • 82 ' Zi ( 1 ) If the impedance Z x is a pure capacitance (Ci), and furthermore gi = g 2 = g, the virtual impedance (Z 2 ) is an inductance: Z 2 =ja> C, gl ' 82 which can also be written as: 1-2 C, g 2 ' ( 2 ) (2a) If a second capacitor (C 2 ) is connected across the second set of terminals, the result is a parallel (LC) tuned circuit. If gi = g 2 = g> Ci = C 2 = C and the input and output impedances (R ( and R 2 ) are equal, the resonant frequency (f Q ) and quality factor (Q) are: , _ ° 2jtC (3) Q = ig-R. (4b) Without further calculation it can be stated that the resonant frequency and quality factor of a series tuned circuit (figure 4) are given by the same formulae. It is obvious from the above that the in- put and output impedances of the ampli- fiers should be as high as possible to obtain a high quality factor. The slope of the amplifiers should also be high if a high quality factor is required; however this leads to a high resonant frequency unless relatively large capacitors are used. A simple calculation shows that for, say, Q = 1000 at f 0 = 100 Hz the gyration constant (or slope) must be g = 2.1 O' 3 (if the input and output impedances in parallel are taken to be 1 Mf2) and capacitors Ci = C 2 35 30/iF are needed. If these capacitors are electrolytics the equivalent leakage resistance may exceed the value assumed above (1 MS2), so that a still higher value for g and hence for the capacitors is required, and so on. . . Having explained the theory of the gyrator, we can now discuss some prac- tical circuits. One tun gyrator This particular circuit is used fairly regu- larly, although it is doubtful whether many people realise that it works as a gyrator! The basic circuit is shown in figure 5a, and figure 5b shows the same circuit with more theoretical symbols. It is clear- ly an asymmetrical gyrator: the transistor is the inverting amplifier, with a gyration constant: gi =- = S = 40I C . v i The collector-to-base resistor (R b ) is the non-inverting “amplifier”, with a gyra- tion constant: The second approximation is based on the assumption that v 2 is far greater than Vj , which is usually the case in practice. The impedance conversion is therefore defined in this case as: z i ~ Rb 2 gl -82 • Zi ~S* zr A practical application of this gyrator is shown in figure 6. In this case the slope is approximately equal to S = 40I C =» 200 • 10” 3 A/V, so that the virtual impedance of the section inside the dotted lines is ap- proximately: R b 47 • 10 3 Z * • Z, asjW ’ 2 • 10 2 ssjco • 250. In other words, it behaves like a coil with an inductance of 250 H! Adding the capacitor (C 2 ) across the output gives a low-pass filter with a cut-off frequency of approximately 0.3 Hz. This means that it is a very useful smoothing circuit for the power supply of a preamplifier, for instance. The quality factor is very low, of course - theoretically Q « 1 in this case! - so that it is usually unsuitable for other applications. The minidrum gyrator The gyrator used in the minidrum (else- where in this issue) is a rather more complicated circuit; see figure 7. The inverting amplifier in this circuit is T 9 . The collector load impedance for this transistor is a current source (T g ), 6 7 so that the gyration constant is simply: _ i 2 _ 1 Sl V! R 21 +rd+r e ’ in which rj and r e are the dynamic (= AC) resistances of the diode (D 5 ) and the emitter of T 9 respectively. These im- pedances are determined by the current through T g and T 9 , which is approxi- mately 0.2mA, so that re * rd * 45i * 125n ’ The non-inverting amplifier consists of the long-tailed pair T 5 /T 7 (a differential amplifier), with the current source T 6 as collector load impedance for T 7 . The gyration constant (g 2 ) is determined in this case by R 13 , r e (T s ) and r e (T 7 ), which have approximately the same values as R 21 , rj and r e (T 9 ) in the above formulae. From this it follows that g 2 *=gi =g* 1.4 • lO" 4 , so that a capacitor of 220/iF across one pair of terminals will be “gyrated” into an inductance of 10,000 H across the other set of terminals! This particular gyrator circuit has some outstanding characteristics. In the first place it is symmetrical (g 2 = g 2 = g), as shown above; furthermore the DC bal- ance is maintained over a wide range of supply voltages without any adjustment, and the current consumption is low (approximately 2mA with a 6 V supply). Finally, the performance is mainly de- termined by the closeness in value of the (nominally) 6k8 resistors to one another. This means that if all resistors are, say, 5% too high in value (i.e. all are 7k 1) the performance does not deteriorate. In the minidrum (bassdrum) capacitors C[ and C 2 are added, so that a resonant circuit is obtained; when the circuit is excited by a pulse which is applied to the input marked P it delivers a decaying sine-wave, of which the frequency is: 2ttVCi C 2 32 Hz, see figure 8. The quality factor of the resonant circuit itself depends on the current gain of the transistors used, and can vary between about 60 and 200. However, in the mini- drum an extra damping resistor (R 23 ) is added which brings the Q down to approximately 10. It is interesting to note that in this particular gyrator circuit the collector of T 5 can be used as a fairly low impedance output without influencing the quality factor, and the base of T 7 can be used as an input. Because these two points are in phase, a resistor of, say, 100k connected between them will cause the circuit to oscillate. In effect this is an LC oscillator, of which the frequency is determined by C t and C 2 . improved 7segment display As discussed elsewhere in this issue, it is possible to distinctly improve the read- ability of 7-segment numeric displays by adding an extra stroke at the top of a six and at the bottom of a nine. The other circuit uses two transistors to achieve this; however, in digital circuits one often has a few gates “left over” because the integrated circuits are not fully used. In this case it is more attractive to use these extra gates to achieve the improved readability. The circuit in figure 1 uses two open- collector NAND gates (e.g. SN7401). The output of N t switches “low” when the B and C inputs are “high”, i.e. for a 6 and for a 7, and adds the stroke at the top of the 6. The 7 uses this stroke anyway, so it remains unchanged. In the same way N 2 switches on the bottom stroke when A and D are “high”, i.e. for the 9. The circuit in figure 2 uses one NAND gate (either normal or open collector, e.g. SN7400orSN7401) and two diodes. As soon as the lower stroke (segment d) is turned on, the stroke at the to (segment a) is added via D! . This is t case for a 2, 3, 5, 6, and 8, all of which except the 6 had this stroke anyway. The result is that only the display for the 6 is changed : the stroke at the top ii added. When input D is “high”, i.e. for 8 and 9, N! switches on the top and bottom segments through D t and D 2 . This gives the extra stroke at the bottom of the nine. Figure 1. Improved readability of the 6 and 9. achieved with two open collector NAND gates. Figure 2. The same improvement can be ob- tained with two diodes and one NAND gate. Psychedelic lights elektor february — 259 psychedelic A favourite gimmick in disco- theques is to use flashing lights, which are usually synchronised to the music. An interesting addition is the psychedelic lamp driver, which livens things up a bit by flashing one or more lamps in a continuously changing rhythm. The electrical wiring in, say, a dance hall is usually such that operating the light switch turns on the room lighting to full brightness. This situation remains un- changed until the switch is operated a second time. The consensus of opinion is that something ought to be done about this rather dreary state of affairs. The first idea that comes to mind is to arrange for the room lighting (or some additional ‘spots’) to switch on and off by itself, without requiring a human operator. However a simple regular on-off rhythm quickly becomes rather boring, so we looked for ways and means to vary the flashing rate. One of the several ways of doing this is to modulate the flashing rate - determined by a multivibrator - according to a sine wave function. Another possibility is to modulate the rate with a sawtooth func- tion. Instead of a rhythmic deviation about some central frequency the flash rate now rises steadily from some starting value and then, when the sawtooth reaches its peak, suddenly drops back to the starting value, i It would of course be possible, by invert- ing the modulating signal, to provide a steady slow-down instead of acceleration. i It turns out that the non-inverted sawtooth provides an attractive effect, so that this is what was chosen. The circuit The basis of this circuit is formed by a simple uni-junction oscillator and a volt- age-controlled astable multivibrator. The sawtooth waveform produced by the cir- cuit around UJT T j is not particularly linear — but it doesn’t need to be in this application. What happens is that capacitor Cj is charged via R3 until the voltage is reached at which Ti’s emitter fires. The ensuing breakdown enables Cj to discharge rapidly through T^s junction and the current-lim- iter R 2 . The voltage across Ci therefore approximates a sawtooth wave. The lin- earity could be improved (for possible other applications) by replacing R3 with a current-source. The charging current is then held at a constant level, without the inverted-exponential decay. The periodic time of the voltage across Ci is about 7 seconds. The sawtooth wave- form voltage is applied, via emitter follow- er T2, to the base resistors of the astable multivibrator formed by T3 and T4. As the applied voltage level increases, the multivibrator’s repetition frequency will rise, vice versa. The multivibrator output is taken from the collector of T4 and applied via resistor R9 to the base of T$ . This transistor there- fore switches between cutoff and satura- tion, to produce a better waveform than that at the collector of T4 (which is ‘spoilt’ by C3 ’s charging current flowing through R 8 when T4 is cutoff). Modulation of the astable multivibrator is only possible in this arrangement when its running frequency is several times the repetition frequency of the sawtooth. With the values given the sawtooth varies the flash frequency in the range 2 to 6 Hz. The sharply-switched current through T s is used to control a triac, which switches the tree lights on and off. The triac is there- fore DC-driven. This has the major ad- vantage that the triac turns on close to the zero-crossings of the mains waveform, so that no interference-causing switching peak arises. The triac specified here re- quires about 50mA triggering current so that the collector resistor for T5, Ru, is selected at 220 ohms. Since the triac can switch up to 6 Amps it is possible to flash several lamps at the same time, to a maximum of about a kilo- watt, allowing for the fact that the resist- ance of the lamp-filaments when cold will be lower than their rated (i.e. hot) value. If several lamps are used, it is actually a better solution to provide each with its own flasher - each unit having a different sawtooth frequency. This frequency can be altered by changing the value of R 3 . A lower value increases the frequency, since the heavier current into Ci will charge this up faster. The circuit is supplied via Di , Rio and D3 directly from the raw AC mains. Observe the skull-and-crossbones symbol on the drawing - and make sure of the insulation! H What's new in Soldering Chemicals? Multicore’s R & D Laboratories are still making news-three important new chemicals for electronics manufacturers. MULTICORE PC 26 ROSIN FOAM FLUX A completely new general purpose liquid soldering flux particularly suitable for the automated soldering of all types of printed circuits. PC 26 provides a unique combination of desi rabl e properties. • Complies with U.K. Ministry Flux Speci- fication D.T.D. 599A. • Eliminates “icicles" and “bridging”. • 0.5% max. halide content and yet gives better soldering than non-approved fluxes with high halide contents. • Leaves negligible flux residues so p.c. boards are dry after soldering, can be handled and inspected easily and have better sales appeal. MULTICORE PC 81 SOLVENT CLEANER & FLUX REMOVER A unique blend of polar and non-polar solvents formulated for degreasing electronic hardware prior to soldering as well as for removing rosin flux residues including ionizable activators after soldering. Itsintermediateboilingrangeof 71 to 80“ C and selective solvency make it ideal for vapour degreasing The boiling range of PC. 81 is higher than fluorinated solvents (approx 46°C) and lower than either trichloroethylene (8rC) or per- chloroethylene (121“C). Also its solvency prop- erti es for rosi n flux removal are superior to fluori- nated solvents without in any way affecting most electronic hardware. As a result, PC. 81 solvent will perform its vapour cleaning function longer and more effectively than fluorinated solvents whose vapour condensation ceases at 46“C with a consequent end to flux removal. Solvent evaporation rate is substantially lower than that of the fluorinated solvents, making it more economical to use in open tanks and vapour degreasers. Multicore PC. 81 is a highly stabilized solvent blend, extremely resistant to thermal or chemi- cal breakdown during prolonged heating or as a result of the introduction of activators from the solution of rosin during its working life Its relatively narrow boiling range and high stability make it readily useable again without property changes after distillation. PC. 81 can also be used for cold cleaning and to reinforce ultrasonic cleaning. Even though its toxicity is relatively low, well ventilated areas are required. PC. 81 is expected to be partic- ularly welcome as it is non-flam- mable and non-combustible under the new British “Highly Inflammable Liquid” Regulations. Supplied in one gallon metal cans and 45 gallon steel drums. Specific Gravity (20C) -1.256 Boiling Range 71 -80 C Toxicity (TLV) 340 ppm Residue on Evaporation- les&than 10 ppm MULTICORE PC 54 CONFORMAL COATING Fullv meets the requirements of the new U.K. Standard 59-47/issue 2 and U.S. Spec. MIL-I-46058C, which are becoming mandatory for the protection of many electronics assemblies against adverse environment, contamin- ation and attack by chemicals. PC. 54 is a two-part epoxy resin system which is conveniently mixed in equal parts by volume. It may be applied by dip, spray or brush to either oneor both sidesof p.c. boards and components where it forms a thin tough coating after curing. PC. 54 will dry in 1 hour under normal ambient conditions and develop itsfull properties after several days at room temperature or it may be cured in 2 to 4 hours at 65*C. A glass fibre brush can be used to remove thecoating locally to enable rework and repair. MULTICORE SOLDERS LIMITED Hemel Hempstead, Herts. HP2 7EP Tel ; Hemel Hempstead 3636 Telex 82363 Other Multicore Soldering Chemicals include a complete range of liquid fluxes and the following special chemicals. PC 2 Multicore Tarnish Remover Cleans tarnish from metal surfaces prior to soldering. PC 10A Activated Surface Preservative Applied after pre-cleaning, preserves solder- ability and need not be removed. PC10D A special version of PC lOAfor application by roller coating machines. PC 90 Peeloff Solder Resist and PC 91 Thinners A temporary solder resist for edge-connector contact areas etc. Replaces masking tape. PC 41 and PC 43 Solder Bath dross inhibitors. PC 52 Protective Coating One-part conformal coating. Can be soldered through. PC 70 Thinners Compatible Solvent blend for use with all rosin fluxes, PC 1 0A and PC 52. Please write for Technical Bulletins on your Company’s letterhead for products which interest you to: ERSIN 3 WJKt aULUCK